Introduction
Despite being relatively small in size and made up of highly fragmented and heterogeneous industries, small- and medium-sized enterprises (SMEs) are epitomised as the 'engines of growth' by economists and the ‘key source of dynamism, innovation and flexibility’ in developed and developing economies (The Economist, 2010; Ng, 2016; OECD, 2010). SMEs account for a large share of total enterprises and make significant contributions to real GDP growth, new job creation, and poverty reduction. In fact, most of the large corporations like Apple and Microsoft initially began as an SME and later evolved into a corporate titan. In Malaysia, SMEs account for 97.3% of all enterprises and contributed 36.3% to the GDP (2005: 30.0%), 65.5% to employment (2005: 56.8%) and 17.6% to exports (2010: 16.4%) in 2015 (SME Corporation Malaysia, 2016). There are two definitions commonly referred to in classifying industrial sectors. For the manufacturing sector, SMEs are defined as firms with a sales turnover not exceeding RM50 million or a number of full-time employees not exceeding 200. For the services and other sectors, SMEs are defined as firms with a sales turnover not exceeding RM20 million or a number of full-time employees not exceeding 75 (SME Corporation Malaysia, 2016). Most SMEs are family-owned and owner-managed businesses where the owners are also the managers holding both management and operational roles.
Limitations and future research direction
The self-report questionnaires that were administered to owner-managers of SMEs constitute a limitation for this research as they may not always produce reliable and valid responses due to a single key informant response rather than multiple responses from firms and industries. This means the data for both independent and dependent variables were collected at the same time from the same source. As a result, this presents the potential problem of common method variance, a variance that is attributable to the measurement method rather than to the constructs the measures represent. In addition, the measures of key constructs are perceptual rather than objective and comparable, consequently relying on correlations for partial representation (Dess & Robinson, 1984). Future research should employ a longitudinal research to assess the development of the key constructs.
