Customer knowledge and export expansion

In this section the structural model related to customer knowledge and the speed and success of export expansion is presented. See Figure 19.

In the theoretical discussion it was mentioned that export expansion capabilities of the alliance learning capability and new product development capability arising from customers’ knowledge are positively related to SPEED and SUCCESS of export expansion. The model shows four structural relationships between the ALLEARCAP, SPEED and SUCCESS and between NPDCAP, SPEED and SUCCESS of export expansion.

As previously stated speed was measured in terms of how satisfied firms were in the timely execution of firstly, being ahead of key competitors; secondly, captur-ing the key export market; thirdly, introduccaptur-ing the right product to customers when they needed it and finally, entering into a market at a time when profit mar-gin opportunities were still available. Success had been measured in terms of fi-nancial indicators as well as the firm’s satisfaction with product performance,

managing the timeframe for the overall project, relationships with alliances in the foreign market and the performance of the firm.

The structural path model for hypotheses 3a and 3b and 4a and 4b is shown in Figure 19.

Figure 19. Structural path model for hypotheses 3 and 4 (a and b).

** show significance at < 0.05 level; *** show significance at < 0.01 level The measurement model was tested by examining a) internal consistency b) con-vergent validity and c) discriminant validity. Only on the scale of SUCCESS did the items such as COMPRDT, MKTSHARE, OVRALPER, PERFCOM, PRDTDELV, PRFTMRGN, ROA, ROI and SALESOBJ not load with values ≥ 0.50. The unreliable items were removed from the constructs for further analysis.

The composite reliability for the reflective latent constructs was > 0.7. Thus, all the reflective constructs showed good internal consistency of measures. The val-ues for the AVE demonstrated that the latent variables of NPDCAP captured 53

% of the valid variance from their indicators, whereas the latent variables of AL-LEARCAP captured 40 % of the variance. The latent dependent variables of SPEED and SUCCESS captured 58 % and 40 % of the variance respectively. The individual item loadings, composite reliability of the reflective latent constructs, the AVEs and the cross loadings are shown in Table 25.

Alliance learning

capability Speed

R2=0.24

New product development capability

0.63 **

0.23 ***

Success

R2=0.38 0.35 ***

-0.03

Table 25. Loadings, cross loadings and the composite reliability of the

Two items from the constructs ALLEARCAP and SUCCESS shared loadings on each others’ block, thus indicating that they refer to similar measures for both the constructs. For example, as can be seen from Table 23, the item of ALLIANCE from the construct of SUCCESS also loaded on the alliance learning capability.

Similarly, the item PRTNREXP from the construct of ALLEARCAP also loaded on SUCCESS. Moreover, as in the previous model, the item KYEXPMKT from the construct of SPEED loaded on the construct of SUCCESS. These are shown as bold and in brackets in the loadings in column 3 and 5 of Table 25. However, the rest of the indicators loaded higher on their respective latent variables as com-pared to other blocks. This problem was taken care of by the average correlation values of the latent variables. The AVEs of all the latent variables were larger than the correlations among the latent variables. Thus, the discriminant validity of the model demonstrated significance at the acceptable threshold levels. See Table 26.

Table 26. Inter-construct correlations among the reflective constructs and the AVE squared along the diagonal

Latent variables ALLEARCAP NPDCAP SPEED SUCCESS

ALLEARCAP 0.63

NPDCAP 0.254 0.72

SPEED 0.330 0.419 0.76

SUCCESS 0.521 0.122 0.357 0.63

R2of the latent dependent variables showed that the model accounted for 22 % of the variance in SPEED and 38 % of the variance in SUCCESS. The Q² redun-dancy indexes (column 3 of Table 27) greater than 0, indicated the predictive rel-evance of the model. The total effects f² of the independent latent variables on the dependent variables indicated that ALLEARCAP had medium impact (23%) on SPEED, whereas more than a large effect (63% as compared to the accepted standard of 35%) on SUCCESS of export expansion. NPDCAP on the other hand, had a large effect on SPEED and had no effect on SUCCESS of export ex-pansion. These effects will be later confirmed in testing the individual hypothe-ses. R², Q²and f²are presented in Table 27.

Table 27. R²,Q² and f²for hypotheses 3 and 4 (a and b)

Latent variables R² Q²

f 2

SPEED

f2

SUCCESS

SPEED 0.24 0.056 -

SUCCESS 0.38 0.148

ALLEARCAP - - 0.23 0.63

NPDCAP - - 0.35 -0.03

Similar to the previous structural model, the assessment of the significance of the structural relationships was carried out by examining the beta path coefficients, t-values and the probability of the t-t-values. Furthermore, the confidence interval and t-values for individual path coefficients were assessed. Path coefficients for the sample and resample in this case were also created by the bootstrap.

The first structural relationship between ALLEARCAP and SPEED was posi-tively significant (β= 0.23; t =2.38; p = 0.01). It was captured in terms of hy-pothesis 3a, which stated that the alliance learning capability of an exporting firm

will have a positive relationship to the speed of export expansion. Thus, hypothe-sis 3a was supported at an observed significance level of 99 %. Learning the needs of the customers from alliances rated significantly on the scale. The mechanisms of inter-firm level knowledge sharing of customer-specific and mar-ket-related information through meetings, document sharing and interaction loaded significantly. Regarding the contribution of partnership-related activities in understanding the needs of the customers, market knowledge possessed by for-eign partners was most important for the export firm.

The second structural relationship between ALLEARCAP and SUCCESS was statistically significant (β= 0.63; t=5.34; p=0.001). This relationship was cap-tured in terms of hypothesis 3b, which stated that the alliance learning capability of an exporting firm will have a positive relationship with the success of export expansion. Thus, hypothesis 3b was accepted at an observed significance level of more than 99 %.

The third structural relationship between NPDCAP and SPEED of export expan-sion (β= 0.35; t= 4.154 p < 0.01) was also significant. New product development capability significantly influenced the speed of export expansion when the firms were first involved in the export market. The competitive advantage of the uct in satisfying customer needs which were not satisfied by a competitor’s prod-uct was also significant. This strprod-uctural relationship was captured in terms of hy-pothesis 4a and stated that the new product development capability of an export-ing firm will have a positive relationship to the speed of export expansion. Thus, hypothesis 4a was accepted at the observed significance level of 99 %.

Finally, the fourth structural relationship between NPDCAP and SUCCESS of export expansion (β=-0.03 t= 0.23 p< 0.00) was negative and insignificant. This also confirmed that NPDCAP had no total effect on SUCCESS of export expan-sion. This hypothesis was rejected. The structural relationship in this model was captured in terms of hypothesis 4b and stated that new product development ca-pability of an exporting firm will have a positive relationship to the success of export expansion. As the relationship was insignificant, hypothesis 4b was re-jected.

6.8 Contextual factors and export expansion