Detection of dependence between two groups of variables

Abstract. The data on the binary recession variable and the latent factors extracted from a large set of economic variables are fitted with a multivariate power-normal distribution. A conditional distribution for the recession variable is obtained from the fitted multivariate distribution. The results based on the US economic data show that the 2.5% point of the conditional distribution provides a good indicator for the start of the next recession.

Section: Methods Based On Multivariate Power-normal Distributionmentioning

confidence: 99%

Prediction of the start of next recession using latent factors

Pooi

Soo²,

Pan

2016

“…In order to achieve the condition given by Equation (4), the constant 2 i should be given by [12]. We next define…”

Section: Multivariate Power-normal Mixture Distributionmentioning

confidence: 99%

“…This data is essentially a summarized version of the individual claims data. A model based on Multivariate Power-Normal Mixture (MPNM) distribution [11][12] is proposed for calculating the aggregate claims liabilities. For the dataset considered, the prediction interval based on the distribution for the aggregate claim liabilities is found to have good ability of covering the observed aggregate claim liabilities.…”

Section: Introductionmentioning

confidence: 99%

Prediction of reserves using multivariate power-normal mixture distribution

Ling¹,

Hin²

2016

“…The multivariate time series model proposed by Pooi [5] applies the following power-normal distribution, which was presented by Yeo and Johnson in [10]. The random variable is said to have a power-normal distribution with parameters and if…”

Section: A Time Series Model Based On Multivariate Power-normal Distrmentioning

confidence: 99%

“…The layout of this paper is as follows. In the next section, the multivariate time series given in [5] is briefly introduced and some results on prediction of deaths rates given in [6] and [8] are highlighted. The subsequent section provides the prediction results obtained using the model with stochastic parameters.…”

Section: Introductionmentioning

confidence: 99%

Prediction of mortality rates using a model with stochastic parameters

Tan¹,

Pooi²

2016

Prediction of future mortality rates is crucial to insurance companies because they face longevity risks while providing retirement benefits to a population whose life expectancy is increasing. In the past literature, a time series model based on multivariate power-normal distribution has been applied on mortality data from the United States for the years 1933 till 2000 to forecast the future mortality rates for the years 2001 till 2010. In this paper, a more dynamic approach based on the multivariate time series will be proposed where the model uses stochastic parameters that vary with time. The resulting prediction intervals obtained using the model with stochastic parameters perform better because apart from having good ability in covering the observed future mortality rates, they also tend to have distinctly shorter interval lengths.