Econometric Analysis of Count Data by Rainer Winkelmann

By Rainer Winkelmann

The publication presents an up to date survey of statistical and econometric concepts for the research of count number info, with a spotlight on conditional distribution types. The booklet starts off with a presentation of the benchmark Poisson regression version. replacement versions tackle unobserved heterogeneity, kingdom dependence, selectivity, endogeneity, underreporting, and clustered sampling. trying out and estimation is mentioned. ultimately, purposes are reviewed in numerous fields.

Show description

Read Online or Download Econometric Analysis of Count Data PDF

Best econometrics books

Stochastic Limit Theory: An Introduction for Econometricicans (Advanced Texts in Econometrics)

This significant new econometrics textual content surveys contemporary advancements within the speedily increasing box of asymptotic distribution conception, with a different emphasis at the difficulties of time dependence and heterogeneity. Designed for econometricians and complicated scholars with restricted mathematical education, the publication in actual fact lays out the required math and chance concept and makes use of quite a few examples to make its information worthy and understandable.

Forecasting Non-Stationary Economic Time Series

Economies evolve and are topic to surprising shifts triggered through legislative adjustments, financial coverage, significant discoveries, and political turmoil. Macroeconometric versions are a truly imperfect software for forecasting this hugely complex and altering procedure. Ignoring those components ends up in a large discrepancy among idea and perform.

Economics of Insurance

The speculation of assurance is gifted during this e-book, mentioned from the perspective of the idea of economics of uncertainty. the primary of top rate calculation which the publication makes use of is predicated on fiscal equilibrium concept and differs from the various top class platforms mentioned by means of actuaries. Reinsurance is constructed within the framework of common fiscal equilibrium thought lower than uncertainty.


This can be an excerpt from the 4-volume dictionary of economics, a reference e-book which goals to outline the topic of economics at the present time. 1300 topic entries within the whole paintings disguise the extensive subject matters of monetary concept. This extract concentrates on econometrics.

Additional resources for Econometric Analysis of Count Data

Example text

The resulting estimator won't have the standard properties of a maximum likelihood estimator. Alternatively, one can treat n as a continuous parameter. In this case, derivatives can be taken. Since n! (n-k)! r(n + 1) r(k+1)r(n-k+1) where r(·) denotes the gamma-function and r(n + 1) = n! if n is an integer, this involves computation of the digamma function. Alternatively, direct differentiation can be based on an approximation of the factorial representation using Stirling's formula k! ~ (21T)1/2kk+1/2 exp (_k){1 + 1/12k} In either case, a logical difficulty arises with respect to the possible sample space of the underlying random variable X if n is a continuous non-negative parameter.

The use of this model in regression analysis requires an expression for the mean. 87) The approximation is exact for c = 0 and for c ---* 1. L + b)1-c 1-c bl - c ~--~--------- This generalized count data model has two more parameters than the Poisson model. L can be expressed in terms of covariates, and the parameters of the model can be estimated by maximum likelihood. 4 Katz Family of Distributions Distributions for non-negative integers can be uniquely represented by their recursive probability ratios [(X P X = ~ k) ) = ~ = f(k, B) - 1 Pk-l k = 1,2, ...

Example 3 Let N be the number of spells of illness and X be the number of visits to a doctor in a given spell. Z gives then the total number of visits to a doctor. Example 4 Let X be a degenerate random variable such that P(X = 1) = 1. Then Z = N. The concept of compounding is not restricted to cases where N is a count random variable although this is of main interest here. One popular count data model, the hurdle Poisson model, can be interpreted as a Bernoulli distribution that is generalized by a truncated-at-zero Poisson distribution (Santos Silva and Windmeijer, 1999).

Download PDF sample

Rated 4.96 of 5 – based on 15 votes