Organisation of the study

5 conjecture about its rivals’ reactions causes the firm to change its own actions. The conjectural variation values are also interpreted as the market power index. In contrast, in this dynamic model, the slope of a reaction function is the actual response rather than the conjecture of the response. The market power index is not directly inferred from but is calculated using the value of the slope. The index nests all possible behaviours, from perfectly competitive to perfectly collusive. In order to capture the specific characteristics of each duopolist, the symmetry assumption—that is common in the previous studies—is relaxed. With a symmetry assumption, the responses of both firms will be identical. Therefore, it can only show whether the firms interact cooperatively or non-cooperatively. Without the assumption of symmetry, the slope of a firm’s reaction function will show the firm-specific response. Hence, a more complex pattern of interdependence will be revealed Gollop and Roberts 1979. The relationship between the firms’ degrees of market power and the type of interaction between the firms provides insights into possible sources of the market power. The linear quadratic specification is useful as it allows the estimation of the market power index without using any cost data. Ideally, firm-level data would provide the best information about market power in the industry. However, firm-level data are insufficient, and therefore group data are used as an alternative.

1.4 Organisation of the study

A plethora of approaches to modelling market power are reported in the literature, focusing on different aspects of the problem and using different frameworks. In order to identify the most suitable model to the Indonesian CPO industry, the main features of the industry are explored in Chapter 2. This chapter explains the reasons behind setting the public estates and private companies groups as the dominant players in the industry. The different characteristics of these groups are discussed, highlighting the requirement of an asymmetric duopoly model. In addition, the existence of an intertemporal link is also presented. Universitas Sumatera Utara 6 In Chapter 3, various models of market power are explored. In comparing these models, the chapter focuses on the main factor that determines market power in an oligopolistic duopolistic market, which is the ability of oligopolists duopolists to respond to their rivals’ actions. These models are divided into models using the structure–conduct– performance approach and the new empirical industrial organization NEIO approach. The structure–conduct–performance approach includes various static models, while the NEIO approach includes both static and dynamic models. In addition, the NEIO static models are grouped into the comparative and conjectural variations models, while the NEIO dynamic models are grouped into the repeated-game and the state-space game models. The assumptions employed in each model are discussed to determine the appropriateness of its application in certain cases. Chapter 4 explores the theoretical and empirical models in greater details. The theoretical model provides the foundation of the state-space game framework. Three different strategies—the open-loop, closed-loop and Markovian strategies—widely used in previous studies are discussed. This is followed by a discussion of the Lagrangean multiplier and dynamic programming as the tools for obtaining the optimal solution. The empirical model presents the adjustment cost model with a linear quadratic specification. The relevance of this model for the CPO industry and the importance of its specification for market power estimation are discussed. The reasons for relaxing the symmetry assumption in the Indonesian CPO industry case are highlighted. The model is estimated using a Bayesian technique in order to impose the stability, convexity and market power properties of the model. In the last section of Chapter 4, the estimation procedure is presented. The model is estimated by using annual data for the period of 1968–2003. The estimation results are presented in Chapter 5. Firstly, the results of the demand equation and the adjustment system are presented. The results from the adjustment system provide insights into the impact of positive reaction functions in avoiding the complicated estimation of the asymmetric case. Secondly, given these estimates, the adjustment cost parameter and market power index are calculated. The Monte Carlo numerical integration results are examined, including a possible explanation of the rejection of the stability, convexity and Universitas Sumatera Utara 7 market power index properties. This is followed by a discussion of the accuracy of the estimation. Finally, the type of interaction and the degree of market power exerted by the dominant groups are detailed. Based on these findings, several policy implications of the analysis are considered in Chapter 6. The effectiveness of the subsidy and export tax policies is examined. The role of the public estates and the effectiveness of Law No.51999 as an instrument to suppress market power in this industry are also discussed. Then, some possible alternative policies to address the market power problem are provided. Limitations of the study are outlined, leading to some avenues for further research. Finally, a number of concluding comments are made. Universitas Sumatera Utara 8

Chapter 2 The Indonesian Crude Palm Oil Industry