Paul I. Louangrath
4. RESEARCH RESULTS
4.5 Combined Industry Group Analysis under Fisher-Tippett-Gnedenko Method
The same procedure was repeated for the ten industries as a group. This second stage of the calculation was accomplished by using the mean of each industry’s eight operating quarters as an individual observation; thus, there are ten individual observed values for the group.
According to the literature, Weibull and Fréchet distributed data sets may be analyzed under the Generalized Extreme Value equation known as the Fisher-Tipett-Gnedenko GEV equation (Embrechts et al., 1999). The result of the tail index calculation is reported in Table 4 in a ranked format according to NPL rates in ascending order.
Table 11: Group Tail Index for Ten Industries
Industry Mean NPL
Rj Z Rk Tail Index (j)
Mining Finance Agriculture Pub. Util.
Construction Services Realty Wholesales Pers. Cons, Manufacture
0.18 0.29 1.47 3.60 3.77 8.97 9.42 15.34 27.21 29.64
(10.88) (10.76) (9.45) (7.09) (6.90) (1.13) (0.63) 5.93 19.10 21.79
- - - - - - - 0.60 0.60 0.60
- - - - - - - j 6.02
Fréchet Distribution
As a group, the NPL rates for the ten industries manifest Fréchet distribution. The differences in distribution findings in tables 3 and 4 is consistent with the literature which suggests that Weibull and Fréchet distribution are common and could be generalized by the Fisher-Tippett-Gnedenko equation.
Using the Hill method to verify the type of extreme data distribution, it was confirmed that ˆk T, 6.02. As a group, the result shows that the group’s NPL is Fréchet distribution.
This Fréchet distribution is also known as fat-tailed or Type II extreme value distribution. In order to determine the shape of the Fréchet distribution, the QQ-plot method was used where
Xi and Yi were calculated by equations (19), (20) and (21).
132
According to the Fréchet distribution equation, only items: x are isolated for the QQ-plot determination. There are five items that meets this condition.
Table 12: enerating QQ-Plot for Linear Regression to Obtain Shape Parameter
Item
Xobs F t( ) 1 / (1F t( ))A ln(A) Xln(ln(A)) Yln(Xobs)
1 2 3 4 5
8.97 9.42 5.34 7.21 9.64
0.13 0.31 0.50 0.69 0.87
1.15 1.46 2.00 3.18 7.71
0.14 0.38 0.69 1.16 2.04
(1.97) (0.97) (0.37) 0.14 0.71
2.19 2.24 2.73 3.30 3.39
With known Xi and Yi, it is possible to determine the linear regression equation in the form of Y a bX . In this case, the linear regression equation is Y 3.02 0.51 X. Thus, the shape of the Fréchet is 1/b1/ 0.51 1.97 . The positive shape tells us that there is an increasing trend with respect to time (Weibull, 1951). According to this decision rule, it means that the NPL trend in Thailand is increasing with time.
The industries may be ranked according to the magnitude of risk derived from their respective NPL rate. This ranking is summarized in Figure 1 reading from left to right:
lowest-to-highest risk level. Two industries: personal consumption and manufacturing are considered extreme cases. These two industries are the most risky. The remaining five industries: public utilities, construction, services, realty, and wholesales, are considered within 0.95 confidence interval or tolerance level. There are three industries: agriculture, finance, mining, and other, are considered non-risky sectors.
The risk assessment of the industries must be read with the scale of the failure. The scale for the Fréchet distribution may be obtained through the maximum likelihood method (Abbas & Yincai, 2012). The likelihood function is given by:
( 1)1 1
, exp
n n
n n na i
i i i
L x
x
(40)Abbas and Yincai further suggest that by solving for:
logLn
,
/ 0, the maximum likelihood of the failure level may be obtained, thus:ˆ 1/
ML n
t
(41)
where = shape of the curve determined by 1/slope; n= sample size which accounts for 7 industries meeting the condition x , and t
(1 / Xi) . In this case, the calculation shows that ˆML10.91. This value is the threshold of NPL level beyond which is considered high risk. According to the likelihood method, three industries are qualified as high risk or risky: wholesales (15.34), personal consumption (27.21) and manufacturing133
(29.64). Under the standard score method, only two industries (personal consumption and manufacturing), were identified as risky because 0.95 CI was used. Under the maximum likelihood method, the threshold point seems to have been reduced to 80%. While the standard score method relies on pre-defined confidence interval as the threshold point, the maximum likelihood method relies on the parameters of the Fréchet distribution. This different approach under the two methods explains the different findings in identifying the risky industry. This final risk identification is summarized in Table 11.
Table 13: Identify Risky Industry under Generalized Extreme Value Method
Industry X(obs) a b Y 3.02 0.51 X Threshold Risky
Mining Finance Agriculture Pub. Util.
Construction Services Realty Wholesales Pers. Cons.
Manufacture
0.18 0.29 1.47 3.60 3.77 8.97 9.42 15.34 27.21 29.64
3.02 3.02 3.02 3.02 3.02 3.02 3.02 3.02 3.02 3.02
0.51 0.51 0.51 0.51 0.51 0.51 0.51 0.51 0.51 0.51
3.11 3.17 3.77 4.86 4.94 7.59 7.82 10.84 16.90 18.14
10.91 10.91 10.91 10.91 10.91 10.91 10.91 10.91 10.91 10.91
No No No No No No No No Yes Yes
In the final analysis, there are two industries which are most risky. These industries are personal consumption (16.90) and manufacture (18.14). The threshold risk level used is 10.91 obtained by ˆML . This conclusion may be confirmed by Cohen’s d effect size determination. Effect size is the measure of the strength of the phenomenon (Ken and Preacher, 2012). The Cohen’s d is given by:
1 2
n
X X
d S
(42)
The mean for the first group is X1(27.21 29.64) / 2 17.53 . The mean for the second group is just the value of the threshold or X210.91. The standard deviation comes from the combined sets; thus: Sn 3.87. The resulting effect size is 1.71. This value must be read for its probability value in the probability distribution table where Z1(1.71)0.956 or 95.6%. Under the 0.95 confidence interval, this effect size is statistically significant. Thus, personal consumption and manufacture industries present real risk in NPL. Lending in these two industries are considered significantly risky endeavor.
5. CONCLUSION
The significance of this research lies in the practical application of the tools proposed:
system failure analysis. This research studies NPL under system failure analysis. Thailand is used as a case study. It was found that Thailand’s NPL betas differ across industries. Some manifest increasing trends while decreasing trends. Time series modeling was reviewed and
134
rejected as not feasible for short-term risk management tool. Time series requires larger sample size. Large sample means longer time for data collection. In NPL where intervention is time sensitive, shorter time and sampler sample size requires a new tool. Extreme Value Theory (EVT) was used as an alternative tool for NPL studies. EVT is more appropriate because it provides the means to identify and rank risk among industries. It was found that out of ten industries, two industry’s NPL exceeded the threshold level determined under EVT. This finding has practical implication for policy and decision makers in NPL risk management. Prior to this research, publications in the field focus on the cause and effect of NPL, but never studied NPL as a close system. This research fills that gap. To that end, this research is a contribution to the field.
135 REFERENCES
Abbas, K., & Yincai, T. (2012). Comparison of Estimation Methods for Frechet Distribution with Known Shape. Caspian Journal of Applied Sciences Research, 1(10), 58-64.
Anderson, T.W. and Darling, D.A. (1954). A Test of Goodness-of-Fit. Journal of the American Statistical Association, 49 (268), 765–769.
Anderson, T. W.; Darling, D. A. (1952). Asymptotic theory of certain ‘goodness-of-fit’
criteria based on stochastic processes. Annals of Mathematical Statistics, 23 (2), 193–
212.
Bank of Thailand (2015). NPLs and Loans Outstanding. Retrieved May, 5, 2015, from https://www.bot.or.th/English/Statistics/FinancialInstitutions/StatNPLsOutstanding/P ages/default.aspx
Beck, R.. Jakubik, P., & Piloui, A. (2013). Non-Performing Loans: What Matters in Addition to Business Cycle?. Working Paper Series, No. 1515/February 2013. European Central Bank.
Box, G.P., & Pierce, D.A. (1970). Distribution of the Residual Autocorrelations in Autoregressive-Integrated Moving-Average Time Series Models. Journal of the American Statistical Association, 65 (332), 1509-1526.
Brockwell, P.J., & Davis, R.A. (2002). Introduction to Time Series and Forecasting (2nd ed.). New York: Springer.
Burnham, K. P., & Anderson, D. R. (2002). Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach (2nd ed.). New York: Springer-Verlag.
Caprio, G., & Klingebiel, D. (1996). Bank Insolvencies: Cross Country Experience. World Bank Policy and Research Working Paper 1574 (Washington).
Dash, M., & Kabra, G. (2010). The determinants of non-performing assets in Indian commercial bank: An econometric study. Middle Eastern Finance and Economics, 7, 94-106.
Edward, A. I. (2000). Predicting Financial Distress of Companies. New York University.
Retrieved on June 25th, 2015 from http://pages.stern.nyu.edu/~ealtman/Zscores.pdf Edward, A. I. (2000). Predicting Financial Distress of Companies: Revisiting the Z-Score and
Zeta Models. New York University. Retrieved on June 25th, 2015 from http://pages.stern.nyu.edu/~ealtman/ Zscores.pdf
Embrechts P., Resnick S., & Samorodnitsky F. (1999). Extreme Value Theory as a Risk Management Tool. North American Actuarial Journal, 3(2), 30-41.
Espinoza, R., & Prasad, A. (2010). Nonperforming Loans in the GCC Banking Systems and their Macroeconomic Effects. IMF Working Paper 10/224 (Washington: International Monetary Fund).
Fulmer, J.G., Moon, J.E., Gavin, T.A., & Ervin, M.J. (1984). A bankruptcy classification model for small firms. Journal of commercial Bank Lending, 66(11), 25-37.
Green, S. (2011). Time Series Analysis of Stock Prices Using the Box-Jenkins Approach.
Electronic Theses & Dissertations, 668, 2.
Guo, H., Pohl, E. & Gerokostopoulos, A. (2013). Determining the Right Sample Size for Your Test: Theory and Application. 2013 Reliability and Maintainability Symposium, January, 2013.
Hu, J.L., Li, Y., & Chiu, Y.H. (2004). Ownership and nonperforming loans: Evidence from Taiwan’s banks. The Developing Economies, 42(3), 405-420.
Hart, B.I. (1942). Significance Level for the Mean Square Successive Difference to the Variance. Annals of Mathematical Statistics, 13(4), 445-447.
IMF’s Compilation Guide on Financial Soundness Indicators (2004). Paragraphs 4.84-4.85.
136
Jappelli, T., Pagano M., & Marco, M. (2008). Households’ Indebtedness and Financial Fragility. CSEF Working Paper 208.
Kaminsky, G., & Reinhart, C. (1999). The Twin Crises: the Causes of Banking and Balance of Payments Problems. The American Economic Review, 89(3), 473–500.
Kaplan, D. (2000). Structural Equation Modeling: Foundations and Extensions. SAGE, Advanced Quantitative Techniques in the Social Sciences series, vol. 10. USA: SAGE Publication.
Kelley, K., & Preacher, K.J. (2012). On Effect Size. Psychological Methods, 17(2), 137–152.
Khemraj, T., & Pasha, S. (2009). The determinants of non-performing loans: An econometric case study of Guyana. The Caribbean Centre for Banking and Finance Bi-annual Conference on Banking and Finance.St. Augustine, Trinidad
Kline, R.B. (2011). Principles and Practice of Structural Equation Modeling (3rd ed.). USA:
Guilford Press.
Lai, C.D. (2014). Generalized Weibull Distributions, Springer briefs in Statistics. DOI 10.1007/978-3-642-39106-4-2; chap. 2, 23-75.
Legault J., (1987). C.A. - Score, A Warning System for Small Business Failures. Bilanas, 29- 31.
Lin, J., Keogh, E., Lonardi, S. & Chiu, B. (2003). A symbolic representation of time series, with implications for streaming algorithms. Proceedings of the 8th ACM SIGMOD workshop on Research issues in data mining and knowledge discovery. New York:
ACM Press. Retrieved on July 24, 2015 from
http://dl.acm.org/citation.cfm?doid=882082.882086
Louzis, D.P., Vouldis, A.T., & Metaxas, V.L. (2010). Macroeconomic and bank-specific determinants of nonperforming loans in Greece: a comparative study of mortgage, business and consumer loan portfolios. Bank of Greece Working Paper, 118.
Maggi, B., & Marco, G. (2009). Modeling non performing loans probability in the commercial banking system: efficiency and effectiveness related to credit risk in Italy.
Working Papers: No. 1 (May 10, 2009). Retrieved on July 27, 2015 from http://phdschool-
economics.dse.uniroma1.it/Economia/Publications/papers/maggi1.pdf
Makridakis, S., & Hibon, M. (1995). ARMA Models and the Box-Jenkins Methodology.INSEAD, 95/45/TM, 1-17.
Masood, O., Bellalah, M., Mansour, W. & Teulon, F. (2010). Non-Performing Loans and Credit Managers’ Role: A Comparative Approach from Pakistan and Turkey.
International Journal of Business, 15(3), 348-362.
NIST (2013). Engineering Statistics Hanbook, sect. 8.2.3.4, online edition. Retrieved on May 12, 2015 from http://www.itl.nist.gov/div898/handbook/apr/section2/apr234.htm Nkusu, M. (2011). Nonperforming Loans and Macrofinancial Vulnerabilities in advanced
Economies. IMF Working Paper WP/11/161.
Pesola, J. (2007). Financial fragility, macroeconomic shocks and banks' loan losses: evidence from Europe. Bank of Finland Research Discussion Papers no.15.
Richard, S. J. (2012). A Handbook of Parametric Survival Models for Actuarial Use.
Sacandinavian Actuarial Journal, 2012(4), 233-257.
Rinaldi, L., & Sanchis A, A. (2006). Household debt sustainability: What explains household nonperforming loans? An empirical analysis. ECB Working Paper, n° 570.
Sampath, S. (2005). Sampling Theory and Methods. Alpha Science International. Harrow, UK.
Shumway, R. & Stoffer, D.S. (2011). Time Series Analysis and Its Applications (3rd ed.) New York, USA: Springer.
137
Springate, G.L.V. (1978). Predicting the possibility of failure in a Canadian Firms.
Unpublished MBA Research Project. Canada: Simon Fraser University.
Stuart A., Ord K., &Arnold S. (1999). Kendall's Advanced Theory of Statistics: Volume 2A—
Classical Inference & the Linear Model §20.2.
Tableman, M. & Kim, J.S. (2004). Survival Analysis Using S: Analysis of Time-to-Date Data. Chapman & Hall/CRC. ISBN 1-58488-408-8; p. 9.
Tanizaki, H. (2004). Computational Methods in Statistics and Econometrics. ISBN 0-8247- 48004-2. p. 47.
Thomas, G. B., & Finney, R. L. (1979). Calculus and Analytic Geometry (5th ed.). Addison Wesley Publishing.
Trichet, J.C. (2010). Reflections on the nature of monetary policy non-standard measures and finance theory. Speech by President of the ECB, Opening address at the ECB Central Banking Conference Frankfurt, 18 November 2010. Retrieved on June 26, 2015 from https://www.ecb.europa.eu/press/key/date/2010/html/sp101118.en.html
Weibull, W. (1951). A statistical distribution function of wide applicability. Journal of Applied Mechanics, 18(3), 293–297.
138
AN EXAMINATION OF THE RELATIONSHIPS BETWEEN SELF-PERCEPTIONS, CONSPICUOUS CONSUMPTION, AND SAVING BEHAVIOR
Chanon Toliang
Thammasat University, Thailand E-mail: [email protected]
Pattana Boonchoo*
Thammasat University
2 Prachan Road, Pranakorn, Bangkok 10200, Thailand E-mail: [email protected]
ABSTRACT
This paper seeks to examine (1) the effects of social status and the three major components of self-esteem, performance, appearance, and social self-esteem, on conspicuous consumption and (2) the impact of conspicuous consumption on saving behavior. The relationships substantiated in this study are based mainly on the perception-behavior linkage within social psychology (Chartrand & Bargh, 1999; Dijksterhuis & van Knippenberg, 1998).
The data used to analyze the proposed relationships in this study were collected through an on-line survey, with a final sample size of 300 consumers. The findings show that only social status and the social dimension of self-esteem significantly affect conspicuous consumption.
Surprisingly, we found no relationship between conspicuous consumption and savings. The results are discussed, along with suggestions for future research.
Keywords: Social Status, Self-Esteem, Conspicuous, Consumption, Saving
139
AN EXAMINATION OF THE RELATIONSHIPS BETWEEN SELF-PERCEPTIONS, CONSPICUOUS CONSUMPTION, AND SAVING BEHAVIOR
1. INTRODUCTION
People consume products and services to fulfill the needs in their everyday lives.
Consumers purchase products or services not only because of their functional benefits but also to gain emotional benefits. One consumption phenomenon that can benefit consumers emotionally is consuming luxury products, which allows them to display their wealth and social status to boost their self-concept (Souiden, M’Saad, & Pons, 2011).
Conspicuous consumption is among the many forms of consumption people use to show off their consumption of luxury products. Now, this kind of consumption can be found in consumers within all social classes because consumers seek social approval and to boost their self-esteem (Johansson-Stenman & Martinsson, 2006). People of various social statuses can use conspicuous consumption to show the social class to which they belong. Perceived social status, together with one’s self-esteem, are important factors in determining one’s level of conspicuous consumption. Consumers with low self-esteem are more likely to use conspicuous consumption to maintain their self-esteem and gain social approval (Johansson- Stenman & Martinsson, 2006; Mason, 1999). While this kind of consumption may benefit luxury brand producers, it may have negative consequences in terms of some consumers’
wellbeing because it can lead consumers into deep financial liabilities.
Based on the above-mentioned phenomenon, it is the aim of this study to examine the inter-relationships among some of the key variables that appear to have some strong linkages with conspicuous consumption: social status, self-esteem, and saving behaviors. All of these constructs are linked together based on the general perception-behavior connection within social psychology (Chartrand & Bargh, 1999; Dijksterhuis & Van Knippenberg, 1998).
By substantiating the inter-relationships among these constructs, we hope to gain insights regarding the specific types of self-perceptions that lead to conspicuous consumption. Managerially speaking, this study will benefit brand managers who seek to influence conspicuous consumption among shoppers and buyers in the market by understanding how consumers’ self-perceptions may lead to this particular kind of consumption.
The remainder of this paper is organized as follows. In the next section (Section 2), we provide a review of the literature related to the four key constructs used in this study. The rationale behind our hypothesized relationships is then provided. Section 3 describes the basic characteristics of our data and research methodology. Section 4 presents the findings and discussions. The paper concludes with the implications of this study, along with some suggestions for future research in Section 5.
140
2. LITERATURE REVIEW