POMS Conferences Main Page

Download POMS Conferences Main Page

Preview text

Risk assessment of India automotive enterprises using Bayesian networks
Purpose: Today's enterprises are facing increased level of risks. It is imperative for companies to assess risk continually. Risks modeling is a complex task because of risks events dependencies and hard task of relevant data. The purpose of this paper is to provide a enterprise risk assessment model that is updated continually. Design/Methodology: Enterprise risk assessment model is provided using Bayesian network methodology for assessing enterprise risks. The networks are used to assess business, economic and external risks and assess its impact on net income of the company. Data for enterprise risk assessment was collected from five automotive companies operating in India. Findings: Companies risk profile results show that companies are more vulnerable to economic risks. The methodology can be used for assessing supply risks, or any business initiative risk. Practical implication: Bayesian network methodology provides a very useful risk assessment tool that combines the advantages of both objectivist and subjectivist risk assessment approaches. Managers can accordingly choose risk mitigation plan. Originality/value: This is a novel effort to provide a assessment tool for enterprise risks in automotive industry.
Key Words: Bayesian statistics, Enterprise risk, Risk profile
1. Introduction:
In economic environment of today, conditions remain challenging for many, and risk management retains its position high on every organization’s agenda. Businesses in the west are grappling with the changes brought about by a post-downturn economy. Postrecession, investors seek a thorough risk assessment of the enterprise and the sector before investing in it. In post-downturn economy, Indian and Chinese economies have emerged as the new global growth locomotives (Siddiqui, 2009).
Enterprise risk management has been hotly debated in boardrooms of many companies. Companies realize their better capital value through increased predictability and lower volatility the key factors contributing to shareholder value. In this context enterprise risk management has emerged in a new business trend. Many risk identification/assessment tools developed to enable a management team to identify and assess risks that their organizations are facing. In the recent past, a numbers of the world's most widely respected companies have collapsed. Analysis's have cited equally well known reasons for these collapses like nonviable business models,

greed, incompetent management and lax regulatory environment. One reason that is not often mentioned is the breadth and depth of these companies' approach to risk management (Borison and Hamm, 2010).
The framework that we propose in this paper would help investors as well as managers to analyze the susceptibility of an enterprise in automotive sector to risk factors. The factors chosen and the resulting Bayesian network would be different for different sectors. The Bayesian network for a sector would depend on the interdependencies of these factors. The Auto sector was chosen because productivity in automotive industry in India is substantially higher than other sectors and it has a huge potential for further improvement, which in turn will pull up the competitiveness of entire manufacturing sector (Draft Automotive Mission Plan, 2006-16) The Indian Automotive Industry started its new journey from 1991 with delicensing of the sector and subsequent opening up for 100 percent FDI through automatic route. Since then almost all the global majors have set up their facilities in India taking the production of vehicle from 2 million in 1991 to 15 million in 2011 (SIAM, 2011). The country with its rapidly growing middle class (450 million in 2007, NCAER report), market oriented stable economy, availability of trained manpower at competitive cost, fairly well developed credit and financing facilities and local availability of almost all the raw materials at a competitive cost has offered itself as one of the favorite destination for investment to the auto makers (Draft Automotive Mission Plan, 2006-16). According to the Society of Indian Automobile Manufacturers, annual vehicle sales are projected to increase to 5 million by 2015 and more than 9 million by 2020 and by 2050, the country is expected to top the world in car volumes with approximately 611 million vehicles on the nation's roads.
The second section provides the motivation and need for the framework presented in the paper. The third section takes up discussion on risk, enterprise risk management and risk assessment. In the section 4, the research methodology has been described which includes a discussion on Bayesian networks and assessment model description. These sections are followed by results, analysis of results and discussions of limitations and future research.
2. Risk Literature:
The possibility of something undesirable to happen is referred to as “risk” colloquially (Rowe, 1977). The critical words in the sentence describing the nature of risk are “possibility” and “undesired”. In the literature, there coexist two parallel definitions of risk. Rescher, (1983) defined risk is an uncertain situation with possible negative outcomes. Risk is the potential variation in outcomes. The variation can be either positive (upside risk) or negative (downside risk) (Williams et al., 1998)
Risk in general can be defined as a collection of pairs of likelihood (L) and outcomes (O):

The distribution pattern of likelihood and outcome pairs is called a risk profile (Avyub, 2003). Definitions of risk must also have a time dimension or a specific time horizon (i.e., day, month, year, etc.) and a specific perspective or view that defines the unit of analysis (i.e., boundaries, etc.). The International Organization for Standardization (ISO 2002) defines two of the essential components of risk: losses (along with related amounts) and uncertainty of their occurrence (likelihood of risky event).
2.1 Enterprise Risk Management (ERM): Risk management is an activity identifying existing and threatening risks, estimating their impacts and taking appropriate measures to reduce or hedge the risks (Pausenberger and Nassauer, 2000). The risk management process can be divided into five steps (Suominen, 2000), as shown in Fig. 1.
In recent years, a paradigm shift has occurred regarding the way organizations view risk management. Instead of looking at risk management from a silo-based perspective, the trend is to take a holistic view of risk management. This holistic approach toward managing an organization’s risk is commonly referred to as enterprise risk management (Gordon et al., 2009).
The following definition of ERM provided by the Casualty Actuarial Society Committee on Enterprise Risk Management (2003, p. 8):
“ERM is the discipline by which an organization in an industry assesses, controls, exploits, finances, and monitors risks from all sources for the purpose of increasing the organization’s short- and long term value to its stakeholders.”

Fig. 1: Steps in Risk Management Process (Suominen, 2000)
2.2 Risk Assessment: Risk assessment is one of the steps in a risk management process. The determination of quantitative or qualitative value of risk related to a recognized threat (also called hazard). Quantitative risk assessment requires calculations of two components of risk (R):, the magnitude of the potential loss (L), and the probability (p) that the loss will occur. (Ayyub, 2003). Fundamentally two different views have evolved over the years on how risk should be assessed. The first view is known as objectivist, or frequentist. This approach requires probabilities are obtained from repetitive historical data and it is based on probabilistic risk assessment (PRA). PRA is the name given to systematic and comprehensive methodology used to evaluate risks mostly for complex technological entities. Consequences are expressed numerically and their likelihoods of occurrence are expressed as probabilities or frequencies. The total risk is the expected loss: the sum of the products of the consequences multiplied by their probabilities (Ramana, 2011). It is generally

used for risk assessment of engineering entities such as power plants and airplanes and finds large scale application in safety and reliability engineering. The second view is termed as subjectivist, or Bayesian view. Bayesian considers the expert judgment as a part of risk assessment. A Bayesian takes not only data into account but also expert' judgment about the situation.
Risk assessment consists of an objective evaluation of risk in which assumptions and uncertainties are clearly considered and presented. For audits performed by an outside audit firm, risk assessment is a very crucial stage before accepting an audit engagement. In project management, risk assessment is an integral part of the risk management plan, studying the probability, the impact, and the effect of every known risk on the project, as well as the corrective action to take should that risk occur (Rausand, 2011). Bayesian view is well accepted in some circles, medical, safety and reliability engineering, but it has not penetrated in enterprise risk management arena.
2.3 Network theory and risk assessment: A network is a simplified representation that reduces a system to an abstract structure. Network modeling and studying have already been applied in many areas, including computer, physical, biological, ecological, logistical and social science. Through the studying of these models, we gain insights into the nature of individual components, connections or interactions between those components, as well as the pattern of connections (Newman, Mark, & Oxford, 2010).
Networks of which each edge has a direction from one vertex to another are called directed graphs. The edges are therefore known as directed edges. Directed networks can be cyclic or acyclic. A cyclic directed network is one with a closed loop of edges. An acyclic directed network does not contain such loop. Since a self-edge – an edge connecting a vertex to itself – is considered a cycle, it is therefore absent from any acyclic network. A Bayesian network is an example of an acyclic directed network.
3. Why Bayesian Network:
Bayesian network is more often used to analyze causal relationships between entities. In the business field, Bayesian Networks are a useful tool for a multivariate and integrated analysis of the risks, for their monitoring and for the evaluation of intervention strategies (by decision graph) for their mitigation (Jensen, 2001; Alexander, 2003). The Enterprise risk can be defined as the possibility that something with an impact on the objectives happens, and it is measured in terms of combination of probability of an event (frequency) and of its consequence (impact). To estimate the frequency and the impact distributions historical data as well as expert opinions are

typically used (Cruz, 2004). In this context Bayesian Networks are a useful tool to integrate historical data with those coming from experts which can be qualitative or quantitative (Fanoni, Giudici, and Muratori, 2005). Thus, Bayesian networks suited all the requirements that we had for risk assessment in Indian automotive sector.

4. Risk Factors for Automotive Industry:

From literature on risk management in automotive sector, we were able to list down following 9 factors which could have an impact on net income of an automotive enterprise:

a) High Competition

f) Regulatory Risk

b) Demand Volatility

g) Economic Instability

c) Exchange Rate Risk

h) Access to Credit

d) Raw Material Price

i) Liquidity Shock

e) Supply Chain Disruptions

After analysis of responses of 20 managers from industry, we established that Access to Credit and Liquidity Shocks do not pose a

serious threat to Indian automotive enterprises. The remaining seven factors have been grouped as below and later used to develop the

Bayesian network which is the risk assessment model for this paper.

4.1 Business Risk: I. High Competition: The Automotive industry in India is highly competitive (Ernst and Young Business Risk report, 2010). The Factors affecting competition include product quality and features, the amount of time required for innovation and development, pricing, reliability, safety, fuel economy, customer service and financing terms. The greater the level of competition for sales in an industry, the more valuable an ERM system should be for a firm within that industry. Thus, there should be positive relation between the degree of industry competition confronting a firm and its need for an ERM system (Gordon and Lawerence, 2009). II. Demand Volatility: Demand for vehicles depends to a large extent on social, political and economic conditions in a given market and the introduction of new vehicles and technologies. This sector is highly correlated with the macroeconomic conditions and in India demand depends a lot on oil prices.

4.2 Economic Risk:
I. Exchange Rate risk: Movement in the value of Rupee determines the attractiveness of Indian products overseas and the price of import for domestic consumption. The trend of export and import has been increasing in the industry and thus importance of exchange rate risk has been increasing. Foreign activities are a source of exchange rate risk (Copeland, 2005).
II. Raw material Price: As a result of increase in raw material prices – mainly metal and energy prices – the volatility in the sector has been on a rise in past few years though the overall volatility is still considered medium.
4.3 External Risk: I. Supply Chain Disruptions: Tier 2 stoppage, disasters, supplier financial stress, suppliers’ union issues are some of the external factors that may lead to supply chain disruptions (Lockamy III and McCormack, 2012). The occurrence of any of these events in the major markets from which a firm purchases materials, parts, components and supplies for the manufacture of its products or in which its products are produced, distributed or sold, may result in disruptions and delays in the operations of firm’s business. (Craighead et al., 2007). II. Regulatory risk: The industry is subject to regulations and legislations related to environmental concerns. Import, export tariff, sales and excise duty also effect the prices of the vehicles affecting the firm and industry as a whole. (Oetzel, Bettis, & Zenner., 2000)
III. Economic instability: The Auto sector has a strong positive correlation with macroeconomic factors. Per capita income, employment levels, size of middle class, interest rates are the major economic parameters that affect this industry. Normally, a stable country with low political risk may encourage the foreign investment whereas countries with high political risk and instability may discourage the foreign investment (Copeland, 2005).
5. Research Methodology:
The research started with identification of a list of 9 factors which affect the automotive industry in India based on literature review. This was followed by a survey of 20 managers from the industry. The process resulted in establishment of seven key factors which lead to risk in this sector. The next step was formation of a Bayesian network to analyze the risk of negative changes in net income of a firm. The subsequent sections shall be explaining the whole process in detail.

5.1 Bayesian network: Bayesian network is defined as a graphical model that efficiently encodes the joint probability distribution for a large set of variables. (Pai et al., 2003). In a Bayesian network, nodes represent random variables and edges represent conditional dependencies; nodes which are not connected represent variables which are conditionally independent of each other (Heckerman,1996). Each node is associated with a probability function that takes as input a particular set of values for the node's parent variables and gives the probability of the variable represented by the node (Jensen, 2001; Heckerman,1996)
Thomas Bayes worked with conditional probability theory in the late 1700s to discover a basic law of probability which came to be known as Bayes’ theorem. Bayes’ theorem states that:
The posterior probability is given by the left-hand term of the equation, P(H|E,c). It represents the probability of hypothesis H after considering the effect of evidence E on past experience c. The term P(H|c) is the a-priori probability of H given c alone. Thus, the apriori probability can be viewed as the subjective belief of occurrence of hypothesis H based upon past experience. The likelihood, represented by the term P(E|H,c), gives the probability of the evidence assuming the hypothesis H and the background information c is true. The term P(E|c) is independent of H and is regarded as a normalising or scaling factor (Niedermayer 2003). Thus, Bayesian networks provide a methodology for combining subjective beliefs with available evidence.
Bayesian networks when first introduced where very time consuming as probability distribution had to be assigned manually. With advancement in computational power and development of heuristic search techniques, Bayesian networks have gained popularity and have been used in areas of supply chain risks, medical diagnosis and computational biology.
Pai et al. (2003) were one of the first researchers to analyze risks using Bayesian networks. Their study examined the risk profile associated with a US Department of Defense (DoD) supply chain for trinitrotoluene (TNT). Bayesian networks have also been used to conduct diagnostics (Kauffmann et al., 2002; Kao et al., 2005), cost optimization studies (Narayanan et al., 2005), and flexibility analysis (Wu, 2005; Milner and Kouvelis, 2005) in supply chains.

5.2 Data collection: The risk categories, the risk factors and the risk measure used is given in table 1. The literature review led us to 9 factors that cause risk in Auto industry. Twenty managers of 15 different automotive firms were mailed and asked to review the list prepared and add their remarks. Based on their responses seven factors mentioned above were selected and later incorporated in the Bayesian network.
Subsequent to this, the seven factors were grouped and the Bayesian network was formed. The risk measures shown in the table 1 were used to get a priori probabilities for each of the seven factors for five companies operating in automotive sector in India. The sample calculation for the network has been shown in Appendix-I.

6. Model description: The first step in model development was identification of factors which primarily cause risk in this sector. The seven factors that were arrived at after literature review and responses from managers have been grouped and used to create a Bayesian network which is shown below:

High competition

Demand volatility

Exchange Rate risk

Economic instability

Supply Chain Disruptions

Regulatory Risk

Raw Material price

Business Risk

Economic Risk

External Risk

Impact on Net Income
Figure 2 Bayesian network representing risk factors in Indian automotive sector The Fig. 2 can exists at two states: Yes and No. Causality links are shown in edges through parent nodes to child nodes. For example demand volatility is a parent node to business risk. Child nodes are conditionally dependent on their parent nodes. Thus from Fig.2, it

Preparing to load PDF file. please wait...

0 of 0
POMS Conferences Main Page