Measuring the Competition Index in the Indonesian Manufacturing Industry: The Structure–Conduct–Performance Paradigm

Abstract

This study proposed a fresh measure of the competition index in the Indonesian manufacturing industry during the period of 1990–2015. The competition index was estimated using indicators in the dimensions of the structure–conduct–performance paradigm. This research used principal component analysis to weigh the indicators in a dimension as well as to weigh the dimensions of the structure–conduct–performance in the final estimation of the competition index. This study found that the Indonesian manufacturing industry had a low competition. The dimension of structure had the highest contribution to the competition index compared to the dimensions of conduct and performance. The performance was also positively affected by the structure and conduct variables. This research contributed to the literature by providing a new competition index using the complete aggregated dimensions of structure, conduct, and performance. Policymakers could also use this competition index to indicate the level of competition in the industry as well as the sources of competition.

1. Introduction

Competition measure is an important indicator for policymakers and anti-trust authorities. This measure provides an indication on whether the market is competitive. This measure can also be a signal on whether the firms in the market have the potential to conduct anti-competitive practices. Therefore, this competition measure can help the anti-trust authorities to anticipate and mitigate the possibility of the cartel and other competitive practices emerging in the future.

The measure of competition is usually based on a single indicator, such as industrial concentration and price-cost margin (see [1,2,3,4,5]). These single indicators do not provide a comprehensive measure of the competition. For example, the industrial concentration only views the competition based on the market structure. Also, the price-cost margin only captures the competition based on the firm or industrial performance. The criteria of a sole indicator to measure the competition may also not apply across all sectors. For example, the high industrial concentration in a few sectors may not reflect a low competition, since the high industrial concentration can be caused by a few firms in the industry that allocate their resources efficiently [6]. Thus, the single indicator may be misleading in regard to showing the real competitive conditions in the market. Therefore, research involving the construction of a comprehensive approach in estimating the competition measure is important.

Previous research measured the competition using the dimensions of the structure–conduct–performance (SCP) paradigm but were not fully applied. The dimension of structure refers to the industry condition shaped by the composition of the firms and the characteristics of the competition within the industry. The dimension of conduct is defined as the firm’s strategy for improving or maintaining its performance in the industry. Finally, the dimension of performance refers to the achievement of the firm’s or market’s goals that were affected by a competitive environment, including the market structure and firm conduct in the industry. Previous studies have also applied the relationship between the indicators or the dimensions of the SCP to indicate the existence of competition in the market. For example, the authors of [4,7,8,9,10,11,12,13] used the indicators of the market structure and performance as well as their relationships in measuring the existence of market power as an indication of competition in the manufacturing sector. They hypothesized that the high market structure i.e., industrial concentration, could be an indicator of low competition if the effect of the market structure on the price-cost margin or profitability was positive. Also, the low competition existed if the high market structure reduced the technical efficiency [6]. Most of this research found that high market structure caused a high inefficiency (meaning a high price-cost margin or a low technical efficiency), indicating a low competition in the market. Regarding the multi-indicators of the competition measure in the SCP paradigm, Petit [14] estimated a competition index for the cartel detection in the Netherlands (NL) manufacturing industry, but their research also did not fully apply the SCP paradigm in constructing their index. As there must be connections between the SCP dimensions, an incomplete use of the SCP paradigm to evaluate the competition might still result in bias. Here, the bias means that measuring the competition with just one of the SCP dimensions does not fully reflect the actual business competition, since other SCP dimensions may produce different results. Using this new method, all dimensions of the SCP are connected and support each other in measuring the competition through the aggregation process. Therefore, this method can reduce the bias that can be encountered when only using one dimension of the SCP. Thus, estimating the competition index using the full concept of the SCP paradigm is still relevant.

This study provides a new estimate of the competition measure, i.e., the competition index, using more comprehensive indicators in the dimensions of the SCP paradigm. Moreover, each dimension of the SCP was estimated. This study applied principal component analysis (PCA) to weigh each dimension in estimating the competition index. This competition measure was then applied to the Indonesian manufacturing industry. Indonesian manufacturing is an important sector in the Indonesian economy. This industry has been contributing more than 20% to the gross domestic product (GDP) since the 1990s [15]. In spite of this fact, this industry may also be experiencing a market distortion, since the authors of [2,16] found that the industry contained a high industrial concentration, and that it classified the industry into a tight oligopoly structure. Therefore, the Indonesian manufacturing industry is a nice example case for the application of the competition measure.

According to the prior justification, this research could help academics and policymakers by offering a new comprehensive measure of the competition. With this competition index, the issue of partial measures, which might not fit a measure of the competition itself, could be reduced. Competition is also one of the key drivers of competitiveness. Thus, understanding the competition through the competition index will improve the rate of sustainability in the context of the economy. A novel approach that was used in this study in measuring the competition was the use of all the aggregated dimensions of the structure, conduct, and performance. This research also additionally linked the dimensions of the structure, conduct, and performance using an econometrics method.

2. Literature Review

Previous studies have provided the measures of competition. The measures were mostly based on the single indicator or dimension of the SCP paradigm. For example, the authors of [1] provided a new way to measure competition based on the firm’s profit. Their research provided a new measure called the relative profit differences (RPD), which was claimed to be better than the price-cost margin (PCM) in explaining the competition. Their results supported the research of the studies published by the authors of [17,18,19,20], who showed that the PCM increased with more intense competition. Thus, the indicator of the PCM could not fully represent the competition. In spite of this, the PCM was still used as a measure of competition in empirical research, such as the studies published by the authors of [21,22,23,24,25]. Furthermore, Hamid [26] investigated the effect of the market structure on banks’ profits in five ASEAN countries. Their research applied industrial concentration as a measure of competition. Hartley and Belin [4] analyzed the SCP in the defense industry and used the “conduct” dimension as a measure of competition, which also had connections with the structure and performance. Kimani et al. [11] investigated the effect of the competition measured using the market structure on the conduct and performance. Their findings suggested that the market structure and conduct affected the performance. Lastly, the Global Competitiveness Report published by the World Economic Forum [27] also conducted a survey and measured competition by including an indicator of competition as a part of competitiveness. Nevertheless, their indicator was still very general and did not measure the level of competition in the industry.

Furthermore, other studies have measured competition based on the relationship between the market structure and performance. For example, the studies published by the authors of [4,11,12,13,26,28] found that the high industrial concentration could be an indication of low competition as the high industrial concentration positively affected the price-cost margin. Also, Setiawan and Oude Lansink [6] and Setiawan [29] found that a high industrial concentration could decrease the efficiency as well as the productivity. The firms operating in the highly concentrated industry may therefore not have the pressure to be more efficient and productive.

Regarding the measures of competition based on multi-indicators, Petit [14] established a competition index for the cartel detection in the Netherlands (NL) manufacturing industry. Their index consisted of nine indicators grouped into four dimensions of degree of organization, prices, concentration, and dynamics, which were as follows:

(i)

The degree of organization: the number of trade associations;

(ii)

Prices: NL price versus EU prices;

(iii)

Concentration: HHI, the number of firms, and the import rate;

(iv)

Dynamics: the market growth, churn rate, survival rate, and R&D rate.

Their research found that this competition index could be useful for detecting the sectors that might be engaged with anti-competitive behavior. In spite of this, their research did not apply the concept of all the SCP dimensions. Furthermore, the indicators in their model were not grouped into the dimensions of the SCP, but they stood alone. Thus, their research was not able to fully apply the SCP paradigm to its nature.

Table 1. Application of the SCP paradigm in estimating the competition.

Previous Research	Method of Estimating Competition	Application of the SCP Paradigm
Boone [1]	This research used relative profit differences (RPD) to measure the competition which only represented the dimension of performance.	This research missed the application of the dimensions of structure and conduct in the estimation.
Pruteanu-Podpiera et al. [21], Maudos and de Guevara [22], Aghion et al. [23], Nevo [24], and Klette [25]	This research used the price-cost margin as a measure of competition.	This research only applied the single dimension of performance, instead of applying all dimensions of the SCP.
Hartley and Belin [4], Setiawan et al. [10,30], Khan and Chakraborty [12], Lartey et al. [13], Hamid [26], and Setiawan [28]	The high industrial concentration can be an indication of low competition because high industrial concentration positively affected the price-cost margin.	This research only applied the dimension of structure (industrial concentration) as a measure of competition that may affect the performance.
Petit [14]	This research applied four dimensions, i.e., (i) the degree of organization, (ii) prices, (iii) concentration, and (iv) dynamics.	This research was not specific in measuring the competition, but measured the cartel. Thus, this research did not fully apply the SCP paradigm nor the dimensions of SCP as a measure of competition.
Kimani et al. [11]	This research used the respective market structure and conduct as a measure of competition that can affect the performance.	This research still applied partial dimensions of the market structure and conduct to measure competition.
Setiawan and Oude Lansink [6] and Setiawan [29]	This research applied industrial concentration to indicate the competition which affected the efficiency.	This research only applied the dimensions of structure and performance.

summarized the application of the SCP paradigm from the previous research. From Table 1, it can be seen that these previous studies never applied all the aggregated dimensions of the SCP to construct a competition measure, including the competition index. Therefore, this research contributed to the literature by aggregating the dimensions of the SCP in estimating the competition index.

This research used the indicators in the SCP paradigm using the framework outlined by the authors of [31]. The dimension of structure included the Herfindahl–Hirschman index (HHI), net entry, and turnover of the four big firms based on the research conducted by the authors of [28,30,32]. The higher the HHI, the lower the competition in the industry. The competition was hypothesized to be higher if the net entry and turnover of the four big firms increased.

The dimension of conduct was measured using capacity utilization and growth of market share, as also applied by the authors of [33,34]. Capacity utilization and growth of market share were hypothesized to have positive associations with the competition.

Moreover, the dimension of performance was measured using the price-cost margin and productivity following the research from the authors of [10,28,30,33,35]. The price-cost margin had a negative relationship with competition, while the productivity had a positive association with the competition, respectively.

Regarding the relation between the dimensions of SCP, it was hypothesized that the structure and conduct affected the performance positively [31]. The better structure indicated the better competitive environment for the business. Thus, it could improve the performance of the market. Also, the better conduct could stimulate the competitive interaction between firms, which could thereby improve the performance of the market.

3. Modeling Approach

Regarding the multi-indicators of the competition index, this research used the complete blocks or dimensions of the SCP paradigm instead of the ungrouped nine indicators used by the authors of [14]. Applying the dimensions of the SCP would be useful for the competition authority in the policymaking, since the regulation and competition authority are mostly concerned on the three blocks of the SCP. Grouping the indicators into each dimension of the SCP would provide more comprehensive measures of these dimensions.

To measure the competition using the indicators of a dimension in the SCP, this research employed a method from previous research, applying a few indicators in measuring the latent variables. This previous method used a summation of the indicators using the appropriate weights. Standardization was also applied if the unit measures of these indicators were different between them. For example, the authors of [36] conducted principal component analysis for weighing the indicators of the regional and firm competitiveness before summing all the indicators in the competitiveness index.

As shown in Figure 1, this research used four steps in estimating the competition index, which were as follows: (1) identification of the indicators in each dimension of the SCP, (2) standardization/normalization of these indicators, (3) weighing the indicators in each dimension and weighing the dimensions for the aggregation, and (4) score aggregation.

3.1. Identification of the Indicators in Each Dimension of the SCP

This research used the indicators in each dimension of the SCP to measure the competition. These indicators were also chosen based on the data availability in the survey of the manufacturing industry sourced from the Indonesian Bureau of Central Statistics.

Table 2. Dimensions and Indicators of the SCP.

Dimension	Indicator	Definition and Measures	Effect on Competition
Structure	Herfindahl–Hirschman index	The sum of the total squared market shares of all firms in the industry.	(−)
	Net entry	The difference between the number of firms in the industry between two consecutive years.	(+)
	Turnover of four big firms	Percentage of four big firms in the previous year doesn’t still stay in the four big firms in the current year of the industry.	(+)
Conduct	Capacity utilization	Average percentage of capacity used for the production in the industry.	(+)
	Growth of market share	Average growth of market share of the firms in the industry.	(+)
Performance	Price-cost margin	Average ratio between price and marginal cost represented by the formula: $\mathrm{P}\mathrm{C}\mathrm{M}=\frac{value added-cost of labor}{output}$	(−)
	Productivity	Average ratio between the output and number of labor.	(+)

Source: various sources.

provides the indicators of each dimension of the SCP.

3.2. Indicator Standardization/Normalization

This research used two steps of the indicator standardization. First, this research transformed the indicators that naturally had a negative meaning into ones that had a positive meaning. For example, the indicators of industrial concentration and the price-cost margin were transformed into the inverse of these indicators. The Herfindahl–Hirschman index (HHI) was transformed into 1 minus the HHI to indicate that the higher the HHI, the higher the competition. Furthermore, the price-cost margin (PCM) was transformed into 1/PCM to indicate that the higher the PCM, the better the competition.

The standardization of these indicators was applied using the method offered in the Global Competitiveness Index using the min–max standardization [37], which was as follows:

$$Z=\frac{X-X_{\min }}{X_{\max }-X_{\min }}$$

(1)

where Z was the normalized variable, X was the unnormalized variable, X_min was the minimum value of X, and X_max was the maximum value of X. The min–max standardization closed the range of the score to only between 0 and 100, respectively. Following the standardization procedure, the scores of these indicators were determined to be comparable, with a higher Z score indicating a higher level of competition.

3.3. Weighing the Indicators in Each Dimension

Before the scores of the indicators are summed up, there should be a weight applied to each indicator indicating which indicators are important in this context. The weight was estimated through performing principal component analysis (PCA). With respect to PCA, confirmatory factor analysis (CFA) was applied in the PCA assuming that all the indicators involved had a strong relationship with a dimension and that they had weak relationships with the other dimensions. CFA was determined to be pertinent for this study, since the indicators that were hypothesized to form a dimension were either based on the theory or empirical research. CFA had the same basic equation of PCA, which was as follows:

$$\begin{gathered} Z_1 = {\pi }_1 \times F_1 + \dots + {\pi }_{\mathrm{k}} \times F_k + v_1 \\ ⋮ \\ {Z}_n = {\pi }_{\mathrm{n}} \times F_1 + \dots + {\pi }_{\mathrm{n}} \times F_n + v_n \end{gathered}$$

(2)

where n and k showed the number of indicators and the number of dimensions, respectively, π was a loading factor, Z was the indicator, and F was the factor or dimension. Generally, the weighing method combined with PCA was applied in consecutive steps, which were as follows: (i) indicator identification for each factor/dimension, (ii) testing the indicator’s validity using the Bartlett’s test of sphericity, and (iii) estimating the weights for each indicator and dimension using CFA. The indicators were determined to be valid to measure the dimensions if the Bartlett’s test of sphericity formed a significant result indicating the high correlation between the indicator and its dimension.

3.4. Score Summation

The weight attached to each indicator and dimension was important for summing all the indicators in a dimension or summing all dimensions to form a competition index. This research used the additive aggregation method (AAM) to sum all the scores of the indicators in a dimension or sum all the scores of the dimensions in the competition index. The AAM was applied since the variables assessed had continuous scores, which had no significant differences on the interval scores between these variables. The total score of the indicators in a dimension or the total score of the dimensions in a competition index was formed as a weighted sum of all the normalized indicators and dimensions, respectively.

The AAM forms a weighted sum of a few variables with the same standard of unit (see [37]), which is formulated as:

$$\begin{gathered} D_q={\displaystyle \sum _{i=1}^{N}w{I}_i} \\ CI={\sum }_{i=1}^Q{pD}_q \end{gathered}$$

(3)

where D is a dimension of the S–C–P, CI is a competition index, q indexes the dimensions of the S–C–P, w is the weight of indicator, p is the weight of the dimensions, I is the score of the indicator, and i indexes the indicators-1,…,N. The competition index ranks the subsectors in the manufacturing industry based on their scores every year.

To verify the estimation of these SCP variables, this research also applied a panel data regression analysis to estimate the effects of the structure and conduct variables on the performance variable. The relationship between these variables was modeled as follows:

$$Performanc{e}_{it}={\alpha }_i+{\beta }_{1}Structur{e}_{it}+{\beta }_{2}Conduct+{\epsilon }_{it}$$

(4)

Equation (4) was estimated by either the fixed effect model or the random effect model based on the Hausman (1978) [38] test. The multicollinearity, autocorrelation, and heteroscedasticity problems were also assessed in this model to verify the fulfilment of the classical linear regression model (CLRM) assumption. The positive effects of the structure and conduct variables on the performance might indicate that variable construction of the SCP supported the SCP hypothesis. The latter suggested that the better market structure and conduct in supporting the competition, the better the market performance.

4. Data

This research used the data of the manufacturing survey sourced from the Indonesian Bureau of Central Statistics (BPS) for the period of 1990–2015. This research used the subsector-level of the five-digit International Standard Industrial Classification (ISIC) system; although not ISIC codes, the actual codes that were used were derived from the Klasifikasi Baku Lapangan Usaha Indonesia (KBLI) classification. This research was not able to expand the period of the analysis beyond 2015, since the data of the five-digit level ISIC system after 2015 was not able to be provided by the BPS when this research was conducted. Also, this research did not use the data obtained before the year of 1990 as not all variables were complete before that period. There were more than 400 subsectors listed in the Indonesian manufacturing survey, but this research covered only 390 of these subsectors due to the issue of data availability.

Table 3. Variable descriptions obtained from the period of 1990–2015.

Variable	Mean	Std Deviation	Coefficient of Variation
HHI	0.303	0.254	0.839
Net entry	1.296	22.761	17.563
Turn over for four big firms	0.331	0.238	0.719
Capacity utilization (%)	75.849	7.861	0.104
Growth of market share	0.275	0.309	1.124
Price-cost margin	0.181	0.714	3.945
Productivity	0.586	0.124	0.212

Source: the BPS; our own calculations.

shows the variable descriptions used for the competition index measures. The variables had not yet been normalized at this point. In the estimation of the competition index, the variables in Table 3 were either normalized or standardized using Equation (1).

Table 3 shows that the net entry, growth of market share, and price-cost margin were the variables with the highest variability across the time period and subsectors. These variables had a coefficient of variation value of greater than one. The net entry had the highest variability, indicating that there was a dynamic entry and exit of the firms in the Indonesian manufacturing industry. The mean HHI of 0.303 also indicated that the industry was highly concentrated during the time period of 1990–2015. This supports the findings made by the authors of [2], who found a tight oligopoly structure in the Indonesian manufacturing sector. This highly concentrated industry in the manufacturing sector might cause the industry to have a relatively high price-cost margin reaching the average PCM of 0.181 or 18.1%. This supports the SCP paradigm hypothesizing that the higher concentration increases the price-cost margin, since the firms in the industry tend to collude.

5. Results

Table 4. Weights for the indicators and dimensions of the competition index.

Dimension	Indicator	Weight	Bartlett’s Test of Sphericity
Structure (0.381)	Herfindahl–Hirschman index	0.370	p-value = 0.000
	Net entry	0.431
	Turnover of four big firms	0.199
Conduct (0.372)	Capacity utilization	0.500	p-value = 0.000
	Growth of market share	0.500
Performance (0.247)	Price-cost margin	0.500	p-value = 0.000
	Productivity	0.500

Source: the BPS; our own calculations. Note: values in the parentheses denote the weights for the dimensions.

shows the weight for each indicator in the dimension and the weight for each dimension to form a competition index. Bartlett’s test of sphericity (p-value = 0.000 < 5%) indicated that the indicators fitted well with the dimensions, since the p-value of the Bartlett statistic was statistically significant for each dimension at the 5% critical level. Based on the weights, the dimension of structure had the highest weight (0.381), followed by the conduct (0.372) and performance (0.247) dimensions. The lowest weight of the performance dimension might also indicate that the firms in the industry were enjoying a stable market power. This stable market power could also be indicated by the high price-cost margin. This indicated that the structure was the most important dimension in explaining the competition index. Furthermore, the most important indicator in the dimension of structure was the net entry with weight of 0.431, followed by the HHI (0.370) and the turnover of the four big firms (0.199). The dimension of conduct was explained fairly with the indicators of capacity utilization (0.50) and growth of market share (0.50). Price-cost margin (0.50) and productivity (0.50) also fairly explained the dimension of performance.

Table 5. Average scores of the competition index and its component.

Period	Structure	Conduct	Performance	Competition Index
1990–1994	0.486	0.393	0.295	0.404
1995–1999	0.481	0.367	0.409	0.421
2000–2004	0.536	0.383	0.249	0.408
2005–2009	0.614	0.388	0.260	0.442
2010–2015	0.554	0.367	0.357	0.436
1990–2015	0.535	0.379	0.316	0.423

Source: the BPS; our own calculations.

shows the estimation of the competition index and its components, including the structure, conduct, and performance variables. The competition index had an average score of 0.423 during the period of 1990–2015. This indicated that the competition was low in the industry since the score was still far from the maximum of one. The performance had the lowest average score compared to the structure and conduct dimensions. This may support the findings made by the authors of [2,6,28], who found that the Indonesian manufacturing sector and the Indonesian food and beverages industry exhibited a persistence of a high industrial concentration and price-cost margin, along with a low technical efficiency. From Table 5, it is revealed that there was a dynamic trend of the variables of structure, conduct, and performance. The dimension of performance had a significant increase during the period of 1995–1999, which coincided with the period of the over-heating economy during the period of 1995–1996, along with an economic crisis during the period of 1997–1998. Additionally, there was also an improvement observed for the variables of structure, conduct, and performance for the period of 2005–2009 compared to the period of 2000–2004.

The competition index also exhibited an increasing trend from the period of 1990–1994 to the period of 2010–2015, respectively. The increases in the value of the competition index in the periods of 1995–1999 and 2005–2009 were caused by the tremendous increases in the performance and structure dimensions, respectively.

Table 6. Subsectors with the highest competition index and its components.

KBLI	Industry	Competition Index	S	C	P
32905	Coconut fiber	0.542	0.579	0.680	0.287
26220	Computer accessory	0.519	0.580	0.526	0.420
32120	Jewelry imitation	0.499	0.550	0.545	0.356
26710	Photographic equipment	0.495	0.620	0.468	0.346
14302	Embroidery clothes	0.495	0.613	0.424	0.405
33122	Repair special machine	0.489	0.506	0.496	0.454
33190	Repair other equipment	0.488	0.513	0.478	0.464
14111	Finished clothes from textile convection	0.477	0.699	0.363	0.310
20129	Other fertilizer	0.476	0.556	0.490	0.337
31001	Furniture, wood	0.476	0.683	0.375	0.312
28291	Machinery and printing	0.475	0.449	0.518	0.451
32904	Safety equipment	0.473	0.490	0.459	0.469
32909	Other processing industry	0.472	0.618	0.409	0.335
23921	Clay bricks	0.471	0.596	0.411	0.359
16292	Goods, rattan plaits, and bamboo plaits	0.471	0.603	0.433	0.315
32201	Musical instr., trad.	0.469	0.531	0.456	0.398
31002	Furniture, rattan, and bamboo	0.468	0.654	0.385	0.306
13912	Embroidery textile	0.467	0.581	0.421	0.351
16291	Goods and rattan plaits	0.466	0.613	0.406	0.324
23122	Glass prod and technical	0.465	0.535	0.472	0.351

Source: Our own calculations.

shows the 20 subsectors with the highest competition index and its components. Eighteen subsectors of the group displayed the highest average score of the structure compared to their conduct and performance dimensions. Only the subsectors of 32905 (coconut fiber) and 28291 (machinery and printing) had the highest average score of their conduct dimension. Most of the subsectors included in the subsectors with a high competition index were included in the subsectors with medium-and low-tech industries. For example, the subsectors of coconut fiber; other fertilizer; goods, rattan plaits, and bamboo plaits; musical instrument, trad.; furniture, rattan, and bamboo; and goods and rattan plaits were found to be examples of the industry with medium-and low-tech industries. These new firms could easily enter the market with a low barrier to their entry. The potential sources of barrier to their entry could come from the difficulties in technology acquisition and the capital requirement.

Table 7. Subsectors with lowest competition index and its components.

KBLI	Industry	Competition Index	Structure	Conduct	Performance
12012	Cigarettes and other	0.307	0.346	0.253	0.299
10313	Dried fruit and vegetables	0.338	0.423	0.327	0.226
58110	Installation machine and industrial equipment	0.338	0.351	0.368	0.273
26513	Electronic measuring equipment	0.339	0.370	0.344	0.285
33149	Repair electrical equipment	0.341	0.350	0.359	0.303
26210	Computer	0.345	0.336	0.357	0.303
20123	Compound fertilizer and macro primary	0.352	0.427	0.304	0.297
25130	Steam generator	0.358	0.379	0.411	0.247
30200	Train	0.358	0.395	0.402	0.239
10320	Canned fruit and vegetables	0.364	0.377	0.386	0.299
17013	Value paper	0.364	0.362	0.400	0.314
10391	Soybean tempe	0.364	0.403	0.350	0.316
11010	Liquors	0.365	0.417	0.299	0.373
10779	Other food nec	0.366	0.377	0.383	0.306
25200	Gun and ammunition	0.366	0.312	0.404	0.384
10532	Other process of edible ice (not ice cube)	0.367	0.381	0.379	0.301
10722	Brown sugar	0.368	0.351	0.408	0.315
26792	Binocular and optic instrument (not glasses)	0.368	0.385	0.396	0.309
33141	Repair electric motor, generator, and transformer	0.370	0.377	0.432	0.268
26791	Camera and projectors	0.370	0.428	0.424	0.206

Source: Our own calculations.

shows the 20 subsectors with lowest competition index as well as its components. From Table 7, it is revealed that the subsector 12012 (cigarettes and other) was the subsector with the lowest competition index. The 20 subsectors with the lowest competition index had scores of structure, conduct, and performance of less than 0.4 on average. This could be an indication that there was a consistency of low scores in the structure, conduct, and performance dimensions in classifying the subsectors with the lowest competition. Most of the subsectors grouped into the 20 subsectors with the lowest competition index were mostly included in the subsectors with medium-and high-tech industry. For example, the subsectors of cigarettes and other; electronic measuring equipment; installation machine and industrial equipment; steam generator; train; and binocular and optic instrument (not glasses) were all found to be examples of the industry with medium-and low-tech industries. The requirement of high technology could be a barrier to entry for these new firms to enter the market.

Table 8. Results of the regression of the structure and conduct on the performance.

Independent Variable	Performance Variable
Intercept	0.179 (0.009) p-value = 0.000
Structure	0.141 (0.011) p-value = 0.000
Conduct	0.160 (0.017) p-value = 0.000
p-value of Wald-statistics	0.000

Note: Standard errors in parentheses. Source: authors’ calculation.

estimated the effects of the structure and conduct on the performance as suggested by the SCP paradigm. The estimation applied the random effect model, since the Hausman test rejected the null hypothesis that suggested the fixed effect model. The Durbin–Watson test suggested that the null hypothesis of the presence of autocorrelation was rejected at the 5% critical level. Moreover, the model suffered from the heteroscedasticity problem, which was uncovered using the White test of heteroscedasticity. Thus, the estimation applied the White heteroskedasticity-robust standard errors (see [39]). The correlation coefficient between the structure and conduct dimensions was less than 0.6. Thus, there was no problem of multicollinearity in the model. As displayed in Table 8, it is shown how the variables of structure and conduct affected the performance significantly at the 1% critical level. The variables of structure and conduct had p-values = 0.000, respectively. The coefficients of 0.141 and 0.160 for the consecutive structure and conduct variables suggested that the increase of one unit of these variables increased the performance by 0.141 and 0.160 units, respectively. This support is in agreement with previous research which suggested that the better market structure and conduct improved the performance of the industry (see [6,28,30]).

This research found that the measure of the competition, i.e., the competition index, proposed in this research could classify the Indonesian manufacturing into the industry with a low level of competition. This competition index suggested that technology could be a barrier to entry for new firms to penetrate the market or for incumbents to sustain their position in the market. Moreover, the lowest average score of the performance dimension suggested that the market power was exploited by the dominant firms causing the higher price of Indonesian manufacturing products. This implied that the policymaker should design a policy that might limit the exploitation of the market power, such as by facilitating the entry for new firms or imported product to have more alternative products available for the consumers.

To have more insights into the results of the SCP variables construction, this research also found a consistent finding of the positive effects of the structure and conduct variables on the performance, which indicated that the better market performance could be gained from a better market structure and conduct. The finding might be in line with the findings made by the authors of [2], who found that a higher competition increased the market efficiency. The results of this research suggested that the competition authority should keep monitoring the market structure and conduct of the firms in the industry in order to hinder the market from the anti-competitive behavior that can create distortions and losses to the market.

6. Conclusions

This research estimated the competition index using the complete aggregated dimensions of the structure–conduct–performance (SCP) paradigm. This research employed principal component analysis to weigh the indicators in each dimension of the SCP as well as the dimensions in the competition index. The additive aggregation method was applied to sum up all the indicators in each dimension and all dimensions in the competition index.

This research found that all indicators in the dimension fulfilled the validity test confirming the high correlation of the indicators in the dimension as well as the high correlation between the dimensions in the competition index. This research found that the competition index displayed an increasing trend with fluctuations between periods during the time period of 1990–2015. Based on the competition index, this research found that the Indonesian manufacturing sector had a low level of competition. The dimension of structure had the highest average score compared to the average scores of conduct and performance. This research also found a consistent result with the hypothesis suggesting the positive effects of the structure and conduct variables on the performance variable.

However, this research still had a limitation in terms of the indicators used for the dimensions of the SCP due to the unavailability of data from the survey. In terms of this limitation, other methods might also face the same problem if using the same data of the industry manufacturing survey. However, the characteristics of these data might vary between countries. The Indonesian manufacturing survey also had lags in its report as well as the limitation of using the data from the 5-digit level ISIC system after 2015. Thus, this method can be tested using more indicators of the SCP across other sectors, such as the banking and financial sectors, which may have more data availability with more recent data. Moreover, this method can also be applied to estimate the competition index in other subsectors of the manufacturing industry.