• No results found

Chapter 5 Dynamic Panel Data Estimation Results

5.5 The Dynamic Panel Data Estimation: Model and Results

5.5.2 Dynamic Panel Data Estimation: System GMM Results

This section reports the results of the two step robust system GMM estimates of the relationship between IC and firm performance. We apply the two step SGMM instead of one step because Roodman (2006) argues that two step yields a robust covariance matrix with respect to autocorrelation and heteroscedasticity. Another reason is that the two step method produces the Sargan Test (robust Hensen J-Test), which is not available in the one step SGMM estimation. Tables 5.4 and 5.5 and Appendix Tables E1 and E2 present SGMM results for all 15 markets (developed, emerging and frontier), with ROA, ROE, ATO and P/B as independent variables, respectively. Table 5.4 shows IC efficiency in terms of VAIC is positive and significant at the 1% level in 11 markets and at 5% level in two markets. These findings support our basic argument that IC contributes significantly towards the firm performance in developed and emerging markets with the exception of the Netherlands. The significant relationship between IC and firm performance in the Netherland in the OLS estimation could be the result of spurious regression. The findings from the SGMM estimation are consistent with previous VAIC studies, Clarke et al. (2011) for Australia, Vishnu and Kumar Gupta (2014) for India, Chen et al. (2005) for Taiwan, and Ting and Lean (2009) for Malaysia.

Table 5.4 The Dynamic Panel-Data Estimation: the Two Step Robust System GMM Results with ROA as the Dependent Variable

Model 1 Model 2

L.ROA VAIC L.ROA HCE SCE CEE

Developed Economies Australia 0.324* 0.370* 0.257* 0.025 0.825* 0.708* (0.000) (0.005) (0.000) (0.761) (0.000) (0.000) Austria 0.509* 0.290** 0.325* -0.011 1.154* 0.710* (0.000) (0.029) (0.000) (0.893) (0.000) (0.000) Netherlands 0.291** 0.565 0.344* -0.192 0.918* 0.586* (0.027) (0.138) (0.000) (0.203) (0.000) (0.000) Singapore 0.046 0.297* 0.224* 0.029 0.886* 0.589* (0.764) (0.000) (0.003) (0.493) (0.000) (0.000) Sweden 0.247** 0.303* 0.177** 0.199** 0.846* 0.781* (0.022) (0.000) (0.043) (0.016) (0.000) (0.000) Emerging Economies China 0.564* 0.478* 0.653* -0.346** 1.225* 0.561* (0.000) (0.005) (0.000) (0.038) (0.001) (0.000) Malaysia 0.442* 0.326* 0.227* -0.173 1.520* 0.755* (0.000) (0.005) (0.000) (0.149) (0.000) (0.000) Russia 0.646* 0.544* 0.507* -0.148 0.855** 0.423** (0.000) (0.004) (0.001) (0.715) (0.020) (0.041) South Africa 0.397* 0.219** 0.231* -0.010 0.889* 0.522* (0.000) (0.012) (0.001) (0.901) (0.000) (0.000) Turkey 0.225** 0.305* 0.099 0.233* 0.428* 0.533* (0.022) (0.000) (0.216) (0.000) (0.004) (0.000) Frontier Economies Argentina 0.411* 0.200 0.467* -0.360 0.866* 0.361* (0.000) (0.234) (0.000) (0.120) (0.003) (0.000) Nigeria 0.778* 0.333* 0.760* -0.297 0.223 0.193* (0.000) (0.000) (0.000) (0.518) (0.154) (0.001) Pakistan 0.531* 0.538* 0.463* -0.054 1.035* 0.642* (0.000) (0.001) (0.000) (0.282) (0.000) (0.000) Saudi Arabia 0.407* 0.334* 0.391* 0.145* 0.459** 0.651* (0.001) (0.000) (0.000) (0.007) (0.037) (0.000) Ukraine 0.613* 0.801* 0.489* 0.193 0.763* 0.706* (0.000) (0.001) (0.000) (0.240) (0.001) (0.000)

Note: * ** and *** represent significance at 0.01, 0.05 and 0.10, respectively. Control variables and time dummies are included in all specifications.

Table 5.5 The Dynamic Panel-Data Estimation: the Two Step Robust System GMM Results with ROE as the Dependent Variable

Model 1 Model 2

L.ROE VAIC L.ROE HCE SCE CEE

Developed Economies Australia 0.277* 0.455* 0.225* 0.089 0.693* 0.588* (0.000) (0.004) (0.000) (0.391) (0.000) (0.000) Austria 0.281** 0.233*** 0.193** -0.164** 0.853* 0.407* (0.032) (0.092) (0.042) (0.023) (0.000) (0.000) Netherlands 0.300* 0.105 0.240** -0.202 0.801* 0.384* (0.004) (0.441) (0.015) (0.174) (0.000) (0.000) Singapore 0.052 0.347* 0.207* -0.018 1.029* 0.609* (0.701) (0.000) (0.000) (0.692) (0.000) (0.000) Sweden 0.198*** 0.246** 0.157** 0.011 0.908* 0.591* (0.071) (0.011) (0.017) (0.907) (0.000) (0.000) Emerging Economies China 0.396* 0.707* 0.287* -0.175** 1.065* 0.724* (0.000) (0.002) (0.000) (0.033) (0.000) (0.000) Malaysia 0.407* 0.458* 0.206* -0.015 1.290* 0.716* (0.000) (0.000) (0.000) (0.910) (0.000) (0.000) Russia 0.346* 0.699** 0.448* -0.180 1.302* 0.703* (0.000) (0.020) (0.000) (0.380) (0.000) (0.000) South Africa 0.394* 0.209** 0.151** -0.011 0.980* 0.549* (0.000) (0.024) (0.049) (0.878) (0.000) (0.000) Turkey 0.207*** 0.204* 0.042 0.045 0.626* 0.485* (0.052) (0.000) (0.524) (0.445) (0.000) (0.000) Frontier Economies Argentina 0.505* 0.245*** 0.419* -0.220 0.719* 0.483* (0.000) (0.093) (0.000) (0.247) (0.003) (0.000) Nigeria 0.709* -0.167** 0.552* -0.660*** 0.296* -0.208** (0.000) (0.026) (0.000) (0.076) (0.006) (0.037) Pakistan 0.512* 0.321* 0.398* -0.123** 1.236* 0.526* (0.000) (0.000) (0.000) (0.015) (0.000) (0.000) Saudi Arabia 0.447* 0.175* 0.302* -0.008 0.853* 0.716* (0.000) (0.000) (0.000) (0.806) (0.000) (0.000) Ukraine 0.540* 0.703* 0.379* 0.109 0.827* 0.763* (0.007) (0.006) (0.000) (0.455) (0.001) (0.000)

Note: * ** and *** represent significance at 0.01, 0.05 and 0.10, respectively. Control variables and time dummies are included in all specifications.

Source: Author’s calculations

With the exception of Argentina, Table 5.4 shows VAIC is significant and positively related to ROA in frontier markets at the 5% level. The relationship for Argentina is significant in the static models (OLS & FE) but insignificant in SGMM. Considering VAIC is an accurate measure of IC efficiency (further discussed in the next chapter) these findings are in line with the RB theory. The RB theory argues that IC forms a sustainable competitive advantage for the firm. Our findings endorse this theory that IC significantly contributes towards the financial performance of a firm, which can help the firm to yield above average returns. This also confirms the argument of Kolachi and Shah (2013) that IC is important for all types of firm (big or small) in all types of market (developed or underdeveloped). Zéghal and Maaloul (2010) state that firms can yield extra returns and build a competitive advantage from the effective use of their strategic resources such as IC assets. Our findings are consistent with Zéghal and Maaloul (2010)’s argument, which means when IC efficiency increases, a firm’s performance (ROA) also increases.

The individual component (HCE, SCE and CEE) analysis shows that HCE is insignificant in almost all markets (developed, emerging and frontier); the exceptions are one developed (Sweden), two emerging (China and Turkey) and one frontier (Saudi Arabia) market, which show a weak or negative significant relationship with ROA. This relationship between HCE and ROA was positive and significant in previous studies (Young et al., 2009; Clarke et al., 2011; Vishnu & Kumar Gupta, 2014) which are based on static (OLS and FE) estimators. Our results are consistent with previous studies (Rehman et al., 2011; Alipour, 2012; Mehralian et al., 2012) which show a negative or no significant relationship between HCE and ROA. These findings suggest that firms in most markets, regardless of the economic development stage, treat investment in human capital as expenditure. Our findings cannot endorse the Resource Dependency (RD) theory which argues that firms should utilize their available human resources to increase the value creation of the firm.

The basic argument by Pulic (2004), while developing the VAIC model, was that money spent on humans within the firm should be treated as investment instead of expense. He argues that human resources create value for the firm just like other assets such as land and buildings. Therefore, if spending on those tangible assets are investments then spending on human resources should also be treated as long term investments. This is why Pulic (2004) does not include salaries and wages as expenses in calculating value added (VA). This contradictory result (where our findings are differ from theory) gives rise to two possible scenarios. First, it raises doubts on the reliability of the VAIC model to measure the efficiency of individual components (HCE, SCE and CEE) accurately. It is noteworthy that the measurement of the VAIC model in general and its two components, i.e., HCE and SCE, in particular, have been criticised by Ståhle et al. (2011). In the next chapter, we further discuss criticisms of the VAIC model and modify the original VAIC model. Secondly, since the firm’s owners (shareholders) hire and pay employees to act on their behalf, that spending is treated as

expenditure. That is why these investments are recorded on the expense side of conventional accounting statements (income statement). Some pioneers in the IC field such as Edvinsson and Malone (1997), suggest firms produce separate statements for IC assets. We discuss and test the reliability of the VAIC model in next chapter.

Table 5.4 shows SCE and CEE are positive and significantly related to ROA at the 1% and 5% level in 14 markets; the exception is Nigeria for which no significant relationship was found between SCE and ROA. These findings suggest that firms in all types of market accumulate and utilize SC and CE quite efficiently for the value creation. Our findings, in terms of SCE, agree with the OL theory. Njuguna (2009) states that organizational learning is a process whereby a firm can acquire a new wealth of knowledge that can be translated into innovation and protected in the form of unique processes, models and copyrights. Our findings suggest that firms can transform their structural capital resources into innovation that, in turn, increases the firm’s profitability. Our findings in terms of physical capital (CEE) support the general argument that physical assets are vital resources for the firm to create value (Firer & Williams, 2003; Chan, 2009b; Ting & Lean, 2009; Young et al., 2009; Vishnu & Kumar Gupta, 2014).

The analysis extends to another performance measure, i.e., ROE for a robustness check. Table 5.5 shows ROE, used as a performance measure, produces quite similar results to ROA. The relationship between VAIC and ROE is positive and statistically significant at the 1% level in eight markets, 5% level in four markets and 10% level in two markets (Table 5.5), which means IC increases firm profitability (measured in terms of ROE). The individual components of VAIC analysis produces similar results where HCE is insignificant with ROE in 11 markets. SCE and CEE are positive and significantly related to ROE in all 15 markets at the 5% level. Our findings of SGMM estimation are consistent with some studies (Clarke et al., 2011; Kai et al., 2011; Vishnu & Kumar Gupta, 2014) with only VAIC and CEE are positive and significantly related to firm performance in terms of ROE. These studies used static measures (OLS & FE). The findings again endorse RD theory that IC resources contribute significantly toward firm performance in terms of ROE and ROA.

We also extend our analysis to other dimensions of firm performance, i.e., productivity and market measure (ATO & P/B) to test whether there are any differences in the results when performance is measured in terms of productivity or asset utilization (ATO) and market valuation (P/B). Appendix Tables E1 and E2 show the relationships between IC and ATO and P/B, respectively. These results are quite different from ROA and ROE. The results show IC efficiency is neither significantly related to ATO nor to P/B. However, CEE of the individual components is statistically significant (at 10% or less) with ATO and P/B. These results are consistent with previous studies (Firer & Williams, 2003; Kai et

al., 2011; Mehralian et al., 2012; Gigante, 2013) that report that IC is significantly related to ROA and ROE but weakly or not related to either ATO or P/B.

P/B exhibits two unique characteristics. First, the measure is mostly favoured by investors when making investment decisions. Investors are mostly concerned with the physical resources a firm holds; IC resources are least important to them (Firer & Williams, 2003). Another possible explanation could be that since P/B is based on the closing price on the stock exchange, it might not depict the true situation of the market.

The results from SGMM estimation are mostly consistent to those of the static estimators (OLS and FE) with the exception of HCE, which is significantly related to performance measure but insignificant in this study. Before we generalize these results, it is pertinent to mention that like OLS and FE, SGMM estimations are subject to various diagnostic tests. As argued by Baltagi (2008) and Roodman (2006), one should test the reliability of SGMM results through various tests such as autocorrelation, and validity of instruments. The next section reports and discusses diagnostic tests of SGMM estimators.