Empirical Results

The descriptive statistics of the input and outputs variables are shown in Table 3. It indicates that the operating scale disparity exists in Taiwanese IC packing and testing industry, ex. revenue, patents and market value. Correlation coefficients in Table 4 are significantly positive for all input-output variables, satisfying the property of isotonicity.

[Insert Table 3 about here]

[Insert Table 4 about here]

These DMUs are classified with the different types, which are categorized into foreign direct investment (FDI), company status (listed or OTC) and service type (turnkey means process including packing and testing or non-turnkey), for the further analysis. The first category is if the company has the foreign direct investment (FDI). FDI must affect the organizational coordination, integration and strategic planning, which influence the efficiency of the firms with FDI directly. The second category is company status which is divided into listed company and OTC. The government regulations on the requirements of listed companies are more serious than OTC, ex. the established life, capital, profitability and equity spread. It is worthy to analyze if the difference between listed company and OTC will influence the performance. The last category is service type which is divided into turnkey company and non-turnkey company. These non-turnkey companies which only specialize in single service rendered always depend on their mother companies or look for their business partners to forge an alliance. On the contrary, the turnkey companies own more bargain power and operate more independently. This paper also studies if the diversification will influence the efficiency.

The average technical efficiency of KH model and CCZ model with different types is presented in Table 5. The average technical efficiency of KH model is 0.107 and tends to under-evaluate compared to the average technical efficiency of CCZ model, which is 0.517.

Although there is a discrepancy between the average technical efficiency of the two models, the technical efficiency of the two models presents the similar result, i.e., the average technical efficiency of the firms without FDI is higher than the firms with FDI. The company status and the service type have the same situation. The Mann-Whitney U test of the technical efficiency of the two models with different types is presented in Table 6.

There is a significant result between the technical efficiency and the different types in KH model. However, CCZ model presents an opposite result. This result indicates that these

different types will not influence the technical efficiency under CCZ model because the technical efficiency has been achieved through the adjustment of intermediate measures but KH model can’t provide this kind of function.

[Insert Table 5 about here]

[Insert Table 6 about here]

Employing Lingo 11.0 to operate equation (7), the pre-test result indicates that only VATE’s efficiency score is 1 and the other firms are far away the DEA frontier. So we remove VATE from DMUs to re-generate the optimal values of the intermediate measures.

The adjusted intermediate measures are defined as the optimal values deduct the observed values. The optimal values deduct the observed values and then divided by the observed values equals the average adjusted intermediate measures percentages. The descriptive statistics of the adjusted intermediate measures are shown in Table 7. It is obvious that the deviation of the adjusted intermediate measures is very huge, ex. adjusted patents and adjusted non-operating income percentage. This implies that every inefficient DMU has its own adjusted target no matter increase or decrease. The average adjusted revenue and non-operating income percentage with different types are presented in Table 8. The average adjusted patents with different types are presented in Table 9.

[Insert Table 7 about here]

[Insert Table 8 about here]

[Insert Table 9 about here]

The average adjusted revenue percentage of the firms with FDI is higher than the firms without FDI in recent years. This might imply that even if the firms invest in the foreign market, it is not easy to increase the revenue and need to make more efforts than those without FDI. The average adjusted revenue percentage of the firms which are listed is higher than the firms which are OTC. This also might imply that even if the firms are requested by government regulation seriously, it is not easy to increase the revenue and need to make more efforts than the firms which are OTC. The average adjusted revenue percentage of the firms with turnkey service is higher than the firms without turnkey service.

This also might imply that even if the firms have diversification on IC packing and testing business, it is not easy to increase the revenue and need to make more efforts than the firms which have no diversification.

On the other hand, the average adjusted non-operating income percentage of the firms with FDI is lower than the firms without FDI. The similar results happened in the different company status and service type. We might conclude that it is necessary to enhance the non-operating income for these conservative companies and improve the profitability and market value accordingly.

The average adjusted patents percentage of the firms with FDI, categorized with listed and offer turnkey service should slash and the firms which have no FDI, categorized with OTC and offer non-turnkey service should consider increasing. This implies that overmuch patents development might be waste and the firms which have no FDI, categorized with OTC and offer non-turnkey service might be conservative for the patents strategy or R&D investment. Because the ASE’s observed patent is out of the frontier far, we remove ASE from DMUs to re-calculate the optimal values. The average adjusted patents with different types have an obvious change on the firms with FDI, categorized with listed and offer turnkey service, indicating these firms should also increase patents.

We further use the Mann-Whitney U test to investigate the adjusted intermediate measures with different types. Table 10 shows that the revenue adjustment for FDI, company status and service type are significantly positive and the adjusted revenue percentage has the similar result except FDI. All the non-operating income adjustment, the adjusted non-operating income percentage and the different types are significantly negative.

Both the patents adjustment and FDI are only significantly positive. This indicates that the patents development will be inevitable when a company to invest the foreign market directly.

Removing ASE from DMUs and re-calculating the optimal values, the patents adjustment, FDI and service type are significantly positive.

[Insert Table 10 about here]

We further analyze the tendency of the average adjusted of intermediate measures percentages by yearly. Table 11 shows the average adjusted revenue percentage and non-operating income percentage with different types by yearly and the yearly tendency are shown in Figure 4-1, 4-2 and 4-3. The obvious discrepancy of the average adjusted revenue percentage with different types happened in 2008 due to global economic recession and the weak demand (TSIA, 2008). The tendency reflects that the firms without FDI needs to adjust its percentage of increasing revenue higher and quicker than the firm with FDI during the economic downturn. The similar situations happened in company status and service

type.

Figure 5-1, 5-2 and 5-3 show the yearly tendency of the average adjusted non-operating income percentages with different types. The obvious discrepancy of the average adjusted non-operating income for the firms which have no FDI, categorized with OTC and offer non-turnkey service happened in 2008 and 2009. This result indicates that the firms which have no FDI, categorized with OTC and offer non-turnkey service needs to adjust their percentage of increasing non-operating income higher and quicker during the economic downturn. On the contrary, the firms which have no FDI, categorized with OTC and offer non-turnkey service do not need to adjust their non-operating income over the prosperous period 2006 and 2007.

[Insert Table 11 about here]

[Insert Figure 4-1, 4-2 and 4-3 about here]

[Insert Figure 5-1, 5-2 and 5-3 about here]