Band 217, Heft 1

Under what conditions is concentration as against diversification the better strategy in allocating research effort up to a given capacity limit? We compare the short and the long run and various types of interdependence. This is a theoretical analysis based on a negative exponential relationship between the probability of success and the amount of research input. Zusammenfassung Wann ist Konzentration und wann ist Diversifikation die bessere Strategie beim Einsatz von Forschungstätigkeit? Hier müssen die kurz- und die langfristige Perspektive unterschieden werden. Unsere Analyse beruht auf einem negativ exponentiellen Zusammenhang (nach Vorgabe der Suchtheorie) zwischen Einsatz und Erfolgswahrscheinlichkeit.

In the analysis of time series the autocorrelation or serial correlation coefficient is calculated in a manner not quite analogous to the ordinary correlation coefficient. The difference between these two versions is investigated, mainly in the spirit of descriptive statistics. It emerges, that the version of time series analysis tends to be smaller in absolute value than the ordinary coefficient. For confirmation of this result a threefold approach is taken: In section 3 it is shown, that it holds asymptotically for a large number Ν of observations. In section 4 three simple models of time series are considered, which show the predicted behaviour of the two versions of the autocorrelation coefficient. In section 5 a numerical example is given. Here one can see, that exceptions from the general rule are possible. Therefore emphasis on the asymptotic behaviour is crucial. To consider the problem from a stochastic viewpoint, i.e. the investigation of the sampling behaviour of the two versions seems much more difficult. Therefore in the last section only some results concerning the autocovariance coefficient are derived, which show, that in this case there is no gain in unbiasedness, if the traditional form of the coefficient is used.

Introductory remarks concerning the use of local deterministic trend models in economic time series are followed by explaining the fundamental relations between regression and filter estimates including some considerations on the selection of forecast functions. The general principles of MSE-forecasting are explained and some special methods of MSE-optimizing are discussed.

An international market research institute regularly collects data from doctors about their prescriptions. Due to their increasing importance, computerized doctor practices are becoming a more important sampling element, simultaneously leading to a switched focus from pure individual doctors recruitment to group practice integration. The consequences of a switch from the sampling unit “Doctor” to “Group Practices” is analysed in this paper. Using the data from a real sample, 11 methods of sampling and projection are investigated with regard to precision and their feasibility. Starting with an analysis of variance, the paper ends with a recommendation for best handling such complex situations.

The Boskin Commission recommends to replace the American Consumer Price Index (CPI) (Laspeyres-type) by a cost of living index (COLI). Unfortunately, the Commission did not develop a valid concept of a COLI. Thus it is that a “price index” (an index isolating the price changes) should be preferred. In any case there exists the problem to measure the effects of substitution, if they occur. The usual method of a compromise between base period weighting and current period weighting is not satisfying for several reasons. A possibility little noticed in the literature are indices with an implicit price elasticity of demand of - 1, that is to say indices - arithmetic, harmonic, geometric - weighted with constant values (instead of quantities).

Glasser’s inequality enables the connection of unrelated pairs of measures of disparity and concentration. Using the minimal inequality distribution and the first and second regression through the empirical distribution function, we get new insights in the upper and lower bound of the inequality by Glasser. Using this inequality, new bounds for unrelated measures as Gini Coefficient and Herfindahl Index can be derived. Looking at the graph of these bounds confirms empirical results, that in the region of average concentration R lies above H and in the extremes the opposite holds true.

According to German rent law standard local rents can be determined by means of rent levels. Usually the unweighted mean of rent prices per square meter is used as an indicator. In this paper the question is discussed whether not a weighted average of rent prices is the more adequate indicator.

The Neyman-Pearson theory suffers from two well-known deficiencies: the asymmetric role of size and power and the coarse dichotomic decision a test yields. To overcome these drawbacks, Rüger (1979, 1980) and Wassmer (1993) developed Niveau Profile Tests (Profile Tests). In principle, these tests succeeded. For composite hypotheses, however, they do not yield the intended refinement. For one-side hypotheses, even if there are indifference regions between them, those tests are capable of three decisions only: “reject the hypothesis”, “do not reject the hypothesis” and “remain indifferent”. Therefore, a revision seems indicated. The revision aims at constructing tests which are always capable of more decisions, even if the hypotheses are composite. These tests will be called “Cumulative Niveau Profile Tests”. The article contains the method of constructing those tests, criteria for comparison and a definition of optimality. Existence and uniqueness theorems are formulated.

Since 1990 a dramatic fertility decline took place in the former GDR. It caused a reproduction level of not more than 37 % in 1994. Pessimistic auspices for social security, new opportunities of high quality consumption and a quick adaptation of Western fertility patterns by East Germany’s population are the main reasons assigned to this process. There exist also ideas that the adaptation hypothesis is the only relevant one. In this paper arguments are developed why a precise adaptation of West Germany’s fertility level and pattern by the population of the former GDR is not at all certain. Fertility in the new Bundesländer may continue for a longer time to be much lower than in former West Germany.

With Efron’s law-school example the bootstrap method is compared with an alternative method, called doubling. It is shown, that the mean deviation of the estimator is always smaller for the doubling method. Zusammenfassung Eine Alternative zur bootstrap-Methode wird vorgestellt: Das Doppelungsverfahren. An Hand des law-school Beispiels von Efron werden die Verfahren verglichen. Es ergibt sich, daß die mittlere Abweichung des Schätzers mit Doppelung stets kleiner ist als mit bootstrap.

After a short introduction into the theory of unbiased estimating equations for general regression models and of corrected estimating equations for regression models with error ridden covariables, the precision of an approximative approach due to Stefanski is discussed.

Let F θ ( x ) be a family of distributions in ℝ 1 θ a real parameter and ψ(θ) a non-monotone function. Consider the problem of obtaining a (1 - α)-confidence region for ψ. This problem demonstrates the efficiency of the most general method for the construction of confidence-regions: using the duality with families of tests. The great variety of solutions derivable by this procedure can also be shown. As an example, exact (1 - α)-confidence intervals possessing important desirable properties for ψ = |μ| in case of the Ν (μ; 1)-family are constructed.

Some thoughts and details of a scenario for the further development of the official statistics in the Federal Republic of Germany regarding the present state of art and necessities of future development. These considerations follow some examples which were taken from the business of German official statistics today.