Operational Value-at-Risk in Case of Zero-inflated Frequency

Our article explores an underused mathematical analytical methodology in the social sciences. In addition to describing the method and its advantages, we extend a previously reported application of mixed models in a well-known database about corruption in 149 countries. The dataset in the mentioned study included a reasonable amount of zeros (13.19%) in the outcome variable, which is typical of this type of research, as well as quite a bit of social sciences research. In our paper, present detailed guidelines regarding the estimation of models where the data for the outcome variable includes an excess number of zeros, and the dataset has a natural nested structure. We believe our research is not likely to reject the hypothesis favoring the adoption of mixed modeling and the inflation of zeros over the original simpler framework. Instead, our results demonstrate the importance of considering random effects at country levels and the zero-inflated nature of the outcome variable.

show abstract

“…Following Mouatassim et al [62], suppose that N is the number of outcomes and N 0 is the number of zeros in the data. The likelihood becomes:…”

Section: Appendix A2 Negative Binomial Regression Models and Overdispersionmentioning

confidence: 99%

“…As stated in Mouatassim et al [62], the numerical algorithm can be used to find the estimated value of λ. Thus, the estimated value of p logit can then be determined by replacing λ in (A15) or (A16).…”

Section: Appendix A2 Negative Binomial Regression Models and Overdispersionmentioning

confidence: 99%

Zero-Inflated Generalized Linear Mixed Models: A Better Way to Understand Data Relationships

et al. 2021

View full text Add to dashboard Cite

show abstract

“…However, in the business studies sciences, such as finance and accounting, it is less common to find applications (e.g. Affinito and Piazza, 2009;Mouatassim et al, 2012).…”

Section: Econometrical Analysismentioning

confidence: 99%

“…The numerical algorithm can be used to find the estimated value of λ. Thus, the estimated value of p logit can be determined by replacing λ in ( 6) or ( 7) (Mouatassim et al, 2012).…”

Section: Zero-inflated Count Data Modelsmentioning

confidence: 99%

Cross-border branching in the Latin American banking sector

Fávero

Santos

Serra

2018

IJBM

View full text Add to dashboard Cite

Purpose Branching is not the only way for foreign banks to enter a national market, and it is impractical when there are informational and cultural barriers and asymmetries among countries. The purpose of this paper is to analyze the determinants of cross-border branching in the Latin American banking sector, a region with regulatory disparity and political and economic instability, offering elements to a grounded strategic decision. Design/methodology/approach This study uses data from six Latin American countries. To account for the preponderance of zero counts, classes of zero-inflated models are applied (Poisson, negative binomial, and mixed). Model fit indicators obtained from differences between observed and estimated counts are used for comparisons, considering branches in each region established by banks from every other foreign region of the sample. Findings Branching by foreign banks is positively correlated with the population, GDP per capita, household disposable income, and economic freedom score of the host country. The opposite holds for the unemployment rate and entry regulations of the host country. Originality/value Few paper address cross-border banking in emerging economies. This paper analyzes cross-border branching in Latin America in the context of the current financial integration and bank strategy. Econometrically, its pioneering design allows modeling of inflation of zeros, over-dispersion, and the multilevel data structure. This design allowed testing of a novel country-level variable: the host country’s economic freedom score.

show abstract

“…compare poisson model to the zero inflated model and applied to health insurance data set. Mouatassim, Ezzahid and Belasri(2012) analized operational risk to the zero inflated data and assess the impact of ZIP distribution on the operational capital charge. They concluded that the zero inflated Poisson distribution is better fit then Poisson distribution for modeling operational risk frequency.…”

Section: Literature Reviewmentioning

confidence: 99%

An Application of Claim Frequency Data Using Zero Inflated and Hurdle Models in General Insurance

Sarul¹

2015

Pressacademia

View full text Add to dashboard Cite

Modelling is a key issue to get a fair pricing in insurance. Poisson distribution is the basic model for count data when the assumptions of Poisson process are assured. Since the insurers tend not to state the small claims to get the deductibles and no claim discounts, insurance data has more zeros than expected which makes the contradictions of the Poisson process assumptions that is the equality mean and variance value. However, not to take account excess zeros makes the knowledge deficiency to get the better pricing for the portfolio. In this paper we will compare Poisson Models and Zero Inflated Models which account for this fact for claim frequency data. Also we will review the models in use for count data and also compare Hurdle Models as an alternative to Zero Inflated Models. Our results represent that Hurdle Models are better fit than the other models we compare. We used Akaike's Information Criteria(AIC) as model selection measures.

show abstract

Operational Value-at-Risk in Case of Zero-inflated Frequency

Cited by 10 publications

References 11 publications

Zero-Inflated Generalized Linear Mixed Models: A Better Way to Understand Data Relationships

Zero-Inflated Generalized Linear Mixed Models: A Better Way to Understand Data Relationships

Cross-border branching in the Latin American banking sector

An Application of Claim Frequency Data Using Zero Inflated and Hurdle Models in General Insurance

Contact Info

Product

Resources

About