We consider forecasting from age-period-cohort models, as well as from the extended chain-ladder model.  The parameters of these models are known only to be identified up to linear trends.  Forecasts from such models may therefore depend on arbitrary linear trends.  A condition for invariant...

An algorithm suggested by Hendry (1999) for estimation in a regression with more regressors than observations, is analyzed with the purpose of finding an estimator that is robust to outliers and structural breaks.  This estimator is an example of a one-step M-estimator based on Huber's skip...

It has long been known that maximum likelihood estimation in a Poisson model reproduces the chain-ladder technique. We revisit this model. A new canonical parametrisation is proposed to circumvent the inherent identification problem in the parametrisation. The maximum likelihood estimators for...

We undertake a generalization of the cumulative sum of squares (CUSQ) test to the case of non-stationary autoregressive distributed lag models with quite general deterministic time trends. The test may be validly implemented with either ordinary least squares residuals or standardized forecast...

Johansen derived the asymptotic theory for his cointegration rank test statisic for a vector autoregression where the parameters are restricted so the process is integrated of order one. It is investigated to what extent these parameter restrictions are binding. The eigenvalues of Johansen’s...

It is well-known that convergence of Laplace transforms in a neighbourhood of the origin is a sufficient criteria for weak convergence. It is worthwhile to note that the interior of the set of convergence points is convex. This provides a link to exponential family theory. As an example an...

This paper derives the exact distribution of the maximum likelihood estimator of a first-order linear autoregression with an exponential disturbance term. We also show that, even if the process is stationary, the estimator is T-consistent, where T is the sample size. In the unit root case, the...

We consider the identification problem that arises in the age-period-cohort models as well as in the extended chain-ladder model. We propose a canonical parameterization based on the accelerations of the trends in the three factors. This parameterization is exactly identified and eases...