Bayesian computation via empirical likelihood

Approximate Bayesian computation has become an essential tool for the analysis of complex stochastic models when the likelihood function is numerically unavailable. However, the well-established statistical method of empirical likelihood provides another route to such settings that bypasses simulations from the model and the choices of the approximate Bayesian computation parameters (summary statistics, distance, tolerance), while being convergent in the number of observations. Furthermore, bypassing model simulations may lead to significant time savings in complex models, for instance those found in population genetics. The Bayesian computation with empirical likelihood algorithm we develop in this paper also provides an evaluation of its own performance through an associated effective sample size. The method is illustrated using several examples, including estimation of standard distributions, time series, and population genetics models.

Shrinkage empirical likelihood estimator in longitudinal analysis with time-dependent covariates-application to modeling the health of Filipino children

The method of generalized estimating equations (GEE) is a popular tool for analysing longitudinal (panel) data. Often, the covariates collected are time-dependent in nature, for example, age, relapse status, monthly income. When using GEE to analyse longitudinal data with time-dependent covariates, crucial assumptions about the covariates are necessary for valid inferences to be drawn. When those assumptions do not hold or cannot be verified, Pepe and Anderson (1994, Communications in Statistics, Simulations and Computation 23, 939–951) advocated using an independence working correlation assumption in the GEE model as a robust approach. However, using GEE with the independence correlation assumption may lead to significant efficiency loss (Fitzmaurice, 1995, Biometrics 51, 309–317). In this article, we propose a method that extracts additional information from the estimating equations that are excluded by the independence assumption. The method always includes the estimating equations under the independence assumption and the contribution from the remaining estimating equations is weighted according to the likelihood of each equation being a consistent estimating equation and the information it carries. We apply the method to a longitudinal study of the health of a group of Filipino children.

Portfolio efficiency tests with conditioning information using empirical likelihood estimation

We evaluate the use of Generalized Empirical Likelihood (GEL) estimators in portfolios efficiency tests for asset pricing models in the presence of conditional information. Estimators from GEL family presents some optimal statistical properties, such as robustness to misspecification and better properties in finite samples. Unlike GMM, the bias for GEL estimators do not increase as more moment conditions are included, which is expected in conditional efficiency analysis. We found some evidences that estimators from GEL class really performs differently in small samples, where efficiency tests using GEL generate lower estimates compared to tests using the standard approach with GMM. With Monte Carlo experiments we see that GEL has better performance when distortions are present in data, especially under heavy tails and Gaussian shocks.

Empirical likelihood based longitudinal studies

In longitudinal data analysis, our primary interest is in the regression parameters for the marginal expectations of the longitudinal responses; the longitudinal correlation parameters are of secondary interest. The joint likelihood function for longitudinal data is challenging, particularly for correlated discrete outcome data. Marginal modeling approaches such as generalized estimating equations (GEEs) have received much attention in the context of longitudinal regression. These methods are based on the estimates of the first two moments of the data and the working correlation structure. The confidence regions and hypothesis tests are based on the asymptotic normality. The methods are sensitive to misspecification of the variance function and the working correlation structure. Because of such misspecifications, the estimates can be inefficient and inconsistent, and inference may give incorrect results. To overcome this problem, we propose an empirical likelihood (EL) procedure based on a set of estimating equations for the parameter of interest and discuss its characteristics and asymptotic properties. We also provide an algorithm based on EL principles for the estimation of the regression parameters and the construction of a confidence region for the parameter of interest. We extend our approach to variable selection for highdimensional longitudinal data with many covariates. In this situation it is necessary to identify a submodel that adequately represents the data. Including redundant variables may impact the model’s accuracy and efficiency for inference. We propose a penalized empirical likelihood (PEL) variable selection based on GEEs; the variable selection and the estimation of the coefficients are carried out simultaneously. We discuss its characteristics and asymptotic properties, and present an algorithm for optimizing PEL. Simulation studies show that when the model assumptions are correct, our method performs as well as existing methods, and when the model is misspecified, it has clear advantages. We have applied the method to two case examples.

Efficient parameter estimation in longitudinal data analysis using a hybrid GEE method

The method of generalized estimating equations (GEEs) provides consistent estimates of the regression parameters in a marginal regression model for longitudinal data, even when the working correlation model is misspecified (Liang and Zeger, 1986). However, the efficiency of a GEE estimate can be seriously affected by the choice of the working correlation model. This study addresses this problem by proposing a hybrid method that combines multiple GEEs based on different working correlation models, using the empirical likelihood method (Qin and Lawless, 1994). Analyses show that this hybrid method is more efficient than a GEE using a misspecified working correlation model. Furthermore, if one of the working correlation structures correctly models the within-subject correlations, then this hybrid method provides the most efficient parameter estimates. In simulations, the hybrid method's finite-sample performance is superior to a GEE under any of the commonly used working correlation models and is almost fully efficient in all scenarios studied. The hybrid method is illustrated using data from a longitudinal study of the respiratory infection rates in 275 Indonesian children.

Utilizando estimadores entrópicos generalizados na estimação de modelos de apreçamento de ativos

Este trabalho analisa as propriedades de uma nova medida de má especificação de modelos de apreçamento, que está relacionada com o tamanho do ajuste multiplicativo necessário para que o modelo seja corretamente especificado. A partir disso, caracterizamos o parâmetro que minimiza a medida a partir de um programa dual, de solução mais simples. Os estimadores naturais para esse parâmetro pertencem à classe de Generalized Empirical Likelihood. Derivamos as propriedades assintóticas deste estimador sob a hipótese de má especificação. A metodologia é empregada para estudar como se comportam em amostras finitas as estimativas de aversão relativa ao risco em uma economia de desastres quando os estimadores estão associados a nossa medida de má especificação. Nas simulações vemos que em média a aversão ao risco é superestimada, mesmo quando ocorre um número significativo de desastres.

Optimally Combining Censored and Uncensored Datasets

Economists and other social scientists often face situations where they have access to two datasets that they can use but one set of data suffers from censoring or truncation. If the censored sample is much bigger than the uncensored sample, it is common for researchers to use the censored sample alone and attempt to deal with the problem of partial observation in some manner. Alternatively, they simply use only the uncensored sample and ignore the censored one so as to avoid biases. It is rarely the case that researchers use both datasets together, mainly because they lack guidance about how to combine them. In this paper, we develop a tractable semiparametric framework for combining the censored and uncensored datasets so that the resulting estimators are consistent, asymptotically normal, and use all information optimally. When the censored sample, which we refer to as the master sample, is much bigger than the uncensored sample (which we call the refreshment sample), the latter can be thought of as providing identification where it is otherwise absent. In contrast, when the refreshment sample is large and could typically be used alone, our methodology can be interpreted as using information from the censored sample to increase effciency. To illustrate our results in an empirical setting, we show how to estimate the effect of changes in compulsory schooling laws on age at first marriage, a variable that is censored for younger individuals. We also demonstrate how refreshment samples for this application can be created by matching cohort information across census datasets.

Moment Based Inference with Stratified Data

Many datasets used by economists and other social scientists are collected by stratified sampling. The sampling scheme used to collect the data induces a probability distribution on the observed sample that differs from the target or underlying distribution for which inference is to be made. If this effect is not taken into account, subsequent statistical inference can be seriously biased. This paper shows how to do efficient semiparametric inference in moment restriction models when data from the target population is collected by three widely used sampling schemes: variable probability sampling, multinomial sampling, and standard stratified sampling.

Business on-line? An empirical study of factors leading to the adoption of Internet technologies by Australian SMEs

E-commerce technologies such as a website, email and the use of web browsers enables access to large amounts of information, facilitates communication and provides niche companies with an effective mechanism for competing with larger organisations world-wide. However recent literature has shown Australian SMEs have been slow in the uptake of these technologies. The aim of this research was to determine which factors were important in impacting on small firms' decision making in respect of information technology and e-commerce adoption. Findings indicate that generally the more a firm was concerned about its competitive position such a firm was likely to develop a web site. Moreover the 'Industry and Skill Demands' dimension suggested that as the formal education of the owner/manager increased, coupled with the likelihood that the firm was in the transport and storage or communication services industries, and realising the cost of IT adoption was in effect an investment, then such a firm would be inclined to develop a web site.

Empirical Bayes method in the study of traffic safety via heterogeneous negative multinomial model

In the study of traffic safety, expected crash frequencies across sites are generally estimated via the negative binomial model, assuming time invariant safety. Since the time invariant safety assumption may be invalid, Hauer (1997) proposed a modified empirical Bayes (EB) method. Despite the modification, no attempts have been made to examine the generalisable form of the marginal distribution resulting from the modified EB framework. Because the hyper-parameters needed to apply the modified EB method are not readily available, an assessment is lacking on how accurately the modified EB method estimates safety in the presence of the time variant safety and regression-to-the-mean (RTM) effects. This study derives the closed form marginal distribution, and reveals that the marginal distribution in the modified EB method is equivalent to the negative multinomial (NM) distribution, which is essentially the same as the likelihood function used in the random effects Poisson model. As a result, this study shows that the gamma posterior distribution from the multivariate Poisson-gamma mixture can be estimated using the NM model or the random effects Poisson model. This study also shows that the estimation errors from the modified EB method are systematically smaller than those from the comparison group method by simultaneously accounting for the RTM and time variant safety effects. Hence, the modified EB method via the NM model is a generalisable method for estimating safety in the presence of the time variant safety and the RTM effects.

Empirical analysis on relationship between traffic conditions and crash occurrences

This paper investigates relationship between traffic conditions and the crash occurrence likelihood (COL) using the I-880 data. To remedy the data limitations and the methodological shortcomings suffered by previous studies, a multiresolution data processing method is proposed and implemented, upon which binary logistic models were developed. The major findings of this paper are: 1) traffic conditions have significant impacts on COL at the study site; Specifically, COL in a congested (transitioning) traffic flow is about 6 (1.6) times of that in a free flow condition; 2)Speed variance alone is not sufficient to capture traffic dynamics’ impact on COL; a traffic chaos indicator that integrates speed, speed variance, and flow is proposed and shows a promising performance; 3) Models based on aggregated data shall be interpreted with caution. Generally, conclusions obtained from such models shall not be generalized to individual vehicles (drivers) without further evidences using high-resolution data and it is dubious to either claim or disclaim speed kills based on aggregated data.

Factors Driving Stock Option Grants - Empirical Evidence from Finland

This study contributes to the executive stock option literature by looking at factors driving the introduction of such a compensation form on a firm level. Using a discrete decision model I test the explanatory power of several agency theory based variables and find strong support for predictability of the form of executive compensation. Ownership concentration and liquidity are found to have a significant negative effect on the probability of stock option adoption. Furtermore, I find evidence of CEO ownership, institutional ownership, investment intensity, and historical market return having a significant and a positive relationship to the likelihood of adopting a executive stock option program.

Discriminative learning of Bayesian networks via factorized conditional log-likelihood

We propose an efficient and parameter-free scoring criterion, the factorized conditional log-likelihood (ˆfCLL), for learning Bayesian network classifiers. The proposed score is an approximation of the conditional log-likelihood criterion. The approximation is devised in order to guarantee decomposability over the network structure, as well as efficient estimation of the optimal parameters, achieving the same time and space complexity as the traditional log-likelihood scoring criterion. The resulting criterion has an information-theoretic interpretation based on interaction information, which exhibits its discriminative nature. To evaluate the performance of the proposed criterion, we present an empirical comparison with state-of-the-art classifiers. Results on a large suite of benchmark data sets from the UCI repository show that ˆfCLL-trained classifiers achieve at least as good accuracy as the best compared classifiers, using significantly less computational resources.

Empirical Analysis of Sampling Based Estimators for Evaluating RBMs

The Restricted Boltzmann Machines (RBM) can be used either as classifiers or as generative models. The quality of the generative RBM is measured through the average log-likelihood on test data. Due to the high computational complexity of evaluating the partition function, exact calculation of test log-likelihood is very difficult. In recent years some estimation methods are suggested for approximate computation of test log-likelihood. In this paper we present an empirical comparison of the main estimation methods, namely, the AIS algorithm for estimating the partition function, the CSL method for directly estimating the log-likelihood, and the RAISE algorithm that combines these two ideas.

Urbanization effects on welfare and income diversification strategies of peri-urban farm households in Tigray, Northern Ethiopia: An empirical analysis

Urban areas in many developing countries are expanding rapidly by incorporating nearby subsistence farming communities. This has a direct effect on the consumption and production behaviours of the farm households but empirical evidence is sparse. This thesis investigated the effects of rapid urbanization and the associated policies on welfare of subsistence farm households in peri-urban areas using a panel dataset from Tigray, Ethiopia. The study revealed a number of important issues emerging with the rapid urban expansion. Firstly, private asset holdings and consumption expenditure of farm households, that have been incorporated into urban administration, has decreased. Secondly, factors that influence the farm households’ welfare and vulnerability depend on the administration they belong to, urban or rural. Gender and literacy of the household head have significant roles for the urban farm households to fall back into and/or move out of poverty. However, livestock holding and share of farm income are the most important factors for rural households. Thirdly, the study discloses that farming continues to be important source of income and income diversification is the principal strategy. Participation in nonfarm employment is less for farm households in urban than rural areas. Adult labour, size of the local market and past experience in the nonfarm sector improves the likelihood of engaging in skilled nonfarm employment opportunities. But money, given as compensation for the land taken away, is not crucial for the household to engage in better paying nonfarm employments. Production behaviour of the better-off farm households is the same, regardless of the administration they belong to. However, the urban poor participate less in nonfarm employment compared to the rural poor. These findings signify the gradual development of urban-induced poverty in peri-urban areas. In the case of labour poor households, introducing urban safety net programmes could improve asset productivity and provide further protection.