On two-step estimation for varying coefficient models

In this note we discuss two-step kernel estimation of varying coefficient regression models that have a common smoothing variable. The method allows one to use different bandwidths for different coefficient functions. We consider local polynomial fitting and present explicit formulas for the asymptotic biases and variances of the estimators.     

The Moran coefficient for non-normal data

This paper summarizes findings that extend statistical distribution properties of the Moran coefficient index measuring spatial autocorrelation to non-normal random variables. Pitman–Koopmans theorem results are extended for the mean and the variance of this index. This summary includes a corollary to this theorem, as well as a new theorem (with its proof) and two conjectures implied by it. The first of these statements is supported by asymptotic heuristics; the second is supported by simulation experiment results. Mixture random variables that include heteroscedasticity or overdispersion also are explored. In addition, a simple asymptotic variance for the Moran coefficient is presented, assessed, and found to be very precise for sample sizes as small as 25–100. The principal conclusion is that independence and sample size are the most relevant properties for Pitman–Koopmans theorem results to be extended to non-normal random variables. The independent and identically distributed property reduces the necessary sample size for this extension, as do the properties of symmetry and normal approximation.     

Using restricted CCA for cross-language information retrieval

Canonical correlation analysis is a method of correlating linear relationship between two sets of variables. When not any linear combination of variables is allowed, restricted canonical correlation analysis is appropriate. The method was implemented with alternating least-squares and applied to the cross-language information retrieval on a dataset with officially translated and aligned documents in eight European languages.     

Testing and Valuing Dynamic Correlations for Asset Allocation

We evaluate alternative models of variances and correlations with an economic loss function. We construct portfolios to minimize predicted variance subject to a required return. It is shown that the realized volatility is smallest for the correctly specified covariance matrix for any vector of expected returns. A test of relative performance of two covariance matrices is based on work of Diebold and Mariano. The method is applied to stocks and bonds and then to highly correlated assets. On average, dynamically correct correlations are worth around 60 basis points in annualized terms, but on some days they may be worth hundreds.     

ON AN APPROXIMATE OPTIMALITY CRITERION FOR THE DESIGN OF FIELD EXPERIMENTS UNDER SPATIAL DEPENDENCE

The design of large‐scale field trials where the residuals are correlated has been of recent interest, in large part because of advances in statistical and computational methods of analysis. The construction of designs for correlated data has typically used A‐optimality and is computationally intensive. This involves calculating the inverse of the information matrix for treatments under the supervision of an optimization strategy that explores the design space. We propose an approximation to A‐optimality, using nearest‐neighbour balance, that is less computationally demanding and can achieve at least 95% efficiency relative to A‐optimality in many practical situations.     

A simple derivation of the distribution of the multiple correlation coefficient

It is shown that the non-null distribution of the multiple correlation coefficient may be derived rather easily if the correlated normal variables are defined in a convenient vay. The invariance of the correlation distribution to linear transformations of the variables makes the results generally applicable. The distribution is derived as the well-known mixture of null distributions, and some generalizations when the variables are not normally distributed are indicated.     

An analysis of covariance markov model for interrupted time series experiments

A simple Markov model is described for the analysis of interrupted time series experiments that employ a control series. The procedure involves regressing the experimental preintervention series on the control preintervention series and then testing for change in the experimental postintervention series which is free of change in the control series.     

Estimation of variance of a normal population

In this paper we have suggested two estimators of variance of a normal population developed from the estimators of u² suggested by Govindarajulu and Sahai and Das. These have been shown to be more efficient than the usual estimator s².     

Asymptotic expansions for the moments of serial correlation coefficients

This work is concerned with evaluating the moments of a number of serial correlation coefficients which arise in various ways and where the observations are from the first order autoregressive Gaussian process with known zero mean. The forms considered have biases whose main parts (of order 0(n^-1) , where n is the sample size) are substantially different. They are the intra-class correlation,the maximum likelihood estimators and an estimator whose main part of the bias is sere. The moments are obtained as asymptotic expansions in terms of the parameter of the process and to terms of order 0(n^-3). It is found that removing certain end terms in the denominator of a serial correlation has the effect of reducing the magnitude of the main part of its bias considerably and in one case completely eliminating it. This work extends the results of various authors,e.g.Kandall(1954), Marriott and pope(1954) and white (1961) in the special cases of the first order autogressive process.     

Conditionally gaussian distributions and an application to kalman filtering with stochastic regressors

The work reviews theory of conditionally Gaussian distributions, especially so called theorems on normal correlation. Three theorems are given: the basic, the recursive, and the conditional theorem on normal correlation. They assume that (a,y), (a,x,y), or (a,y,z) has a Gaussian distribution, ussert that (a,y), (a,x,y), and (a,y,z), respectively, are Gaussian, and give formulas for the corresponding conditional mean vectors and variance covariance matrices. A proof is presented for the recursive and the conditional theorem.