An integrated decision support for the energy policies of Turkey
The current account is one of the important financial indicators for the economic credibility of a country, especially for an emerging economy. The main components of the current account are visible balance (trade in goods) and invisible balance (trade in services). Investment incomes and net transfers also affect current account. Basically, a country runs current account deficit if her value of imports is greater than her value of exports, on the other hand, if the value of imports is less than the value of exports the country runs current account surplus.
In January 2014, the current account deficit of Turkey reached $65 billion (7.8% of GDP). As current account deficit is greater than 5% of the GDP, it is expected that the current account deficit will continue to be unsustainable in the coming years. Turkey would have decreased current account deficit had she not rely on foreign energy supplies. An emerging economy needs a lot of energy and Turkey imports a major portion of the oil and the natural gas that she uses.
This research will focus on Turkey’s reliance on foreign energy supplies and examine how convenient energy policies can be created to decrease the current account deficit of the country. For this purpose an integrated decision support approach will be used. Opinions of experts and related issues in the literature will be gathered and a convenient model will be proposed. As the problem on hand is a complex one and consists of several interrelated concepts affecting each other, a network model will be generated. Fuzzy cognitive mapping may be an appropriate approach for this purpose.
Causal cognitive mapping is a method that captures the diverse mental models of the experts in simple directed graphs where concepts are represented by nodes and relations between concepts are represented by an arc from the affecting concept to affected concept. The relation is positive if there is an increase at the affected concept when affecting concept increases. If there is a decrease at the affected concept when affecting concept increases, the relation is negative. By interviews with experts or by examining published reports or studies, the related concepts and interrelations among them can be revealed. These beliefs and judgments are brought together to have an aggregated cognitive map. Qualitative analyses can be conducted on this map. However causal cognitive maps may have drawbacks. For instance if there are two concepts affecting a concept C, and one relation is positive and the other one is negative, it cannot be determined whether C will increase, decrease, or remain same in the long run.
To predict the overall system behavior of the concepts in the cognitive map, a formal analysis can be conducted. One potential approach may be the use of fuzzy cognitive map (FCM) which is based on Fuzzy Set Theory and the Theory of Neural Networks, improves the ability of decision makers to understand the dynamic behavior of causal cognitive maps. Instead of assigning (-1, 0, +1) for the representations of the relations as done in causal cognitive maps, a value at the interval of [-1, +1] is assigned at FCMs. They are therefore fuzzy.
FCM are regarded as a simple form of recursive neural networks, with concepts being the equivalent to neurons. However, concepts of FCMs, are not either off or on (0 or 1), but can take values in-between [0, 1]. Fuzzy concepts are non-linear functions that transform the path-weighted activations directed towards them into a value. When a concept changes its value (a neuron fires), it affects all concepts that are causally dependent upon it. Depending on the sign of the relation and the strength of it, the affected concepts subsequently may change their values as well (further concepts are activated at the network). For this purpose a simulation is conducted as follows: a given state vector with values of -1, 0, or 1 for the concepts (i.e. the value of a concept will be -1 if decision makers let it decrease and it would be 1 if it is let to increase) is multiplied with adjacency matrix (a nxn square matrix representing the fuzzy causal relations among n concept) in each iteration to come up with an updated state vector. System reaches a stable state at the end. By reading the values of concepts at the final state vector, one can understand to which extent which concept will increase and which one will decrease in the long run.
In this research, after making several meetings and interviews with experts and examining the written materials, a FCM will be constructed for examining how convenient energy policies can be created to decrease the current account deficit of the country. Several scenarios can be generated to analyze how the increase or decrease at some certain concepts will affect the overall system behavior of the all concepts of the network.
Prof. Ilker Topcu, Istanbul Technical University
Research Asistant Mine Isik, Istanbul Technical University
Asst. Prof Ozay Ozaydın, Doğuş University