SME Vulnerability Analysis: A Tool for Business Continuity

Abstract

This chapter presents the concept of “vulnerability” and a vulnerability analysis for Small and Medium-sized Enterprises (SMEs) to assess the multiple risks firms might suffer from during economic and financial downturns depending on their management systems and financial health. SMEs can be more flexible and react faster than large firms but, unlike their larger counterparts, most of them do not have at their disposal effective management systems and tools to ensure their sustainability. In a context of economic and financial crisis, an emergency plan for vulnerable SMEs has been carried out in the Basque Country, Spain to address this issue using a vulnerability analysis model.

Annex: Evaluation of the Measurement and the Structural Model

Evaluation of the measurement model

The normal criteria in order to accept an indicator as part of a construct is that it possesses a greater loading than 0.707, which implies that the variance shared between the construct and its indicators is larger than the error variance (Carmines and Zeller 1979). However, some authors believe that this rule should not be so strict and loads of 0.5 or 0.6 can be acceptable in the early stages of scales development (Chin 1998) or when the scales are applied in different contexts (Barclay et al. 1995).

The first results for the 19 original indicators show diverse outcomes. According to Diamantopoulos and Winklhofer (2001), reflective indicators are essentially interchangeable and; therefore, their removal does not change the essential nature of the underlying construct. Given the initial values obtained, some reflective indicators that did not meet the criterion of individual reliability were removed from the model. This situation was not unexpected due to the nature of the model. Without a more in-depth analysis, some of the removed indicators did not seem to be as representative as others and could be somehow redundant. Economic and financial indicators could also have been separated in two constructs following the failure process proposed by Ooghe and De Prijcker (2008).

The tests for the remaining ten indicators show satisfactory results:

Constructs	Indicators	Table 4. Individual item reliability
		Original indicators	Remaining indicators
Structural vulnerability	Strategy	0.7326	0.7166
	Size	0.5920	0.6151
	Continuity	0.0923	Removed
	Management	0.7357	0.7180
	Managers	0.6810	0.6896
	Organization	0.7075	0.7198
Operational vulnerability	Customer concentration	0.3870	Removed
	Internationalization	0.5739	0.6202
	Commercial resources	0.8103	0.8228
	Innovation	0.5023	0.6047
	Production	−0.1808	Removed
	Purchasing	0.1259	Removed
	Quality	0.3497	Removed
Economic-financial vulnerability	Added value	0.0525	0.5110
	Finance	−0.1851	Removed
	Short-term finance	−0.3504	Removed
	Costs	−0.9055	Removed
	Wages	−0.5685	Removed
	Equity	0.3171	0.8360

Composite reliability measures construct reliability. Values starting from 0.7 are accepted in early stages of research but values higher than 0.8 would be preferable (Nunnally 1978). The first and second constructs are more reliable than the third one, which, not by much, but does not meet the standards.

AVE or Average Variance Extracted measures convergence validity or the amount of variance of the construct which is due to its own indicators. Recommended values for AVE should be greater than 0.5 (Fornell and Larcker 1981). The results show that none of the constructs meet the standards, but they are close to them.

Table 5. Construct reliability and convergent validity	Structural vulnerability	Operational vulnerability	Economic-financial vulnerability
Composite Reliability	0.822	0.727	0.636
Average Variance Extracted	0.480	0.476	0.480

When examining discriminant validity for PLS models the accepted method is to show that the square roots of the average variances extracted (diagonal values) are higher than the inter-construct correlations. In this case, the test shows satisfactory results.

Table 6. Discriminant Validity	Structural vulnerability	Operational vulnerability	Economic-financial vulnerability
Structural vulnerability	0.6928
Operational vulnerability	0.550	0.6899
Economic-financial vulnerability	0.211	0.164	0.6928

Evaluation of the structural model

In order to assess the research hypotheses, path-coefficient levels and the contribution of the exogenous constructs to the amount of variance explained in endogenous constructs (R²) have been measured, multiplying path and correlation coefficients. Values starting from 0.2 are accepted in early stages of research but values higher than 0.3 are preferable.

A t-statistic was used to check the significance of path coefficients. Any value greater than 1.6479 is likely to be significant (p < 0.1).

In addition, the predictive power of the model has been tested using the Q2 Stone-Geisser statistic. Cross-validated redundancy (Q) higher than 0 means the model has predictive relevance.

As we can see in Table 7, structural vulnerability is a key element to explain operational vulnerability, but operational vulnerability does not significantly influence economic and financial vulnerability.

Table 7. Structural model evaluation	Path	T-Statistic	Correlation	R2 (amount of variance explained)	Q2 (cross-validated redundancy)
H1: Impact of structural vulnerability on operational vulnerability	0.550	8.3492	0.550	30.25 %	0.0185
H2: Impact of operational vulnerability on economic-financial vulnerability	0.164	1.2645	0.164	2.69 %	−0.4431