This paper analyzes the impact of annual weather fluctuations on the total output of wine and on the share of output of different wine-quality categories in Germany, using a set of wine data from all thirteen German wine regions and daily weather data taken from regional weather stations. The empirical analysis suggests that rising average temperatures have a significantly positive impact on the total output of wine as well as on the output shares of wine in higher-quality categories. The number of freezing days appears to be detrimental to overall production; precipitation during the growing season impairs higher-quality wines in particular. (JEL Classifications: Q21, Q13)

Four models for in vitro embryo production traits in Guzerá cattle were compared: Gaussian (untransformed variable – LIN and transformed in logarithmic scale – LOG), Poisson (POI) and zero-inflated Poisson (ZIP). Data consisted of 5716 ovum pick-up and in vitro fertilization records performed in 1205 cows from distinct regions of Brazil. Analyzed count traits were the number of viable oocytes (NOV), number of grade I oocytes (NGI), number of degenerated oocytes (NDG), number of cleaved embryos (NCLV) and number of viable produced embryos (NEMB). Heritability varied from 0.17 (LIN) to 0.25 (POI) for NOV; 0.08 (LOG) to 0.18 (ZIP) for NGI; 0.12 (LIN) to 0.20 (POI) for NDG; 0.13 (LIN) to 0.19 (POI) for NCLV; 0.10 (LIN) to 0.20 (POI) for NEMB depending on the considered model. The estimated repeatability varied from 0.53 (LOG) to 0.63 (POI) for NOV; 0.22 (LOG) to 0.39 (ZIP) for NGI; 0.29 (LIN) to 0.42 (ZIP) for NDG; 0.42 (LIN) to 0.59 (POI) for NCLV; 0.36 (LIN) to 0.51 (POI) for NEMB. The goodness of fit, measured by deviance information criterion and mean squared residuals, suggested superiority of POI and ZIP over Gaussian models. Estimated breeding values (EBV) obtained by different models were highly correlated, varying from 0.92 for NOV (between LIN-POI) and 0.99 for NGI (between POI-ZIP). The number of coincident animals on the 10% top EBV showed lower similarities. We recommend POI and ZIP models as the most adequate for genetic analysis of in vitro embryo production traits in Guzerá cattle.

This study characterizes the abundance and spatial distribution of siphonophores between the Trinidad Channel (50°06′S) and the Strait of Magellan (52°45′S) in southern Chile, during October–November 2009. Ten species were identified, of which Agalma elegans, Rosacea plicata and Sphaeronectes fragilis are new records for this region. Dominant species showed similar dominance values e.g. Lensia conoidea (26.3%), Dimophyes arctica (24.6%), Lensia meteori (22.2%) and Muggiaea atlantica (20.7%). Eudoxids of L. conoidea and D. arctica represented 97.3% of all eudoxids collected and they were mainly collected in estuarine waters. The highest densities were found in estuarine waters (high vertical stratification and low temperature, salinity and dissolved oxygen values). On the other hand, the lowest densities were found in coastal areas influenced by permanent influx of Sub-Antarctic waters from the Pacific (greater instability and vertical mixing, higher temperatures, salinity and dissolved oxygen values). Temperature and dissolved oxygen were the most important environmental variables. In general, all the dominant species showed a positive association with temperature and a negative association with dissolved oxygen (with the exception of L. meteori). The vertical distribution showed that M. atlantica was mainly distributed in the first 50 m, in association with estuarine waters, while L. conoidea, L. meteori and D. arctica were mainly found in the deeper layer (50–200 m) and in association with modified Sub-Antarctic waters. The comparison of the results obtained in the springs of 1996 and 2009 showed a significant increase in abundance.

The study of the feeding habits of elasmobranch populations should help to elucidate the ecological role of these fishes in the marine ecosystem. Feeding habits of the apron ray, Discopyge tschudii from San Jorge Gulf were investigated. An evaluation of whether the diet of D. tschudii changes with body size, seasons, maturity stage and sex was performed using a multiple-hypothesis modelling approach. Discopyge tschudii preyed mainly on polychaetes. The consumption of polychaetes was higher in males. Also, amphipods and siphons of clams were preyed more heavily in the cold season and Munida gregaria in the warm season. Stomatopods were consumed more with increasing body size. All prey categories were independent of maturity stage. We concluded that D. tschudii is a benthic feeder.

Dobrava-Belgrade virus (DOBV) is the most pathogenic hantavirus in Europe with a case-fatality rate of up to 12%. To detect changes in risk for humans, the prevalence of antibodies to DOBV has been monitored in a population of Apodemus flavicollis in the province of Trento (northern Italy) since 2000, and a sudden increase was observed in 2010. In the 13-year period of this study, 2077 animals were live-trapped and mean hantavirus seroprevalence was 2·7% (s.e. = 0·3%), ranging from 0% (in 2000, 2002 and 2003) to 12·5% (in 2012). Climatic (temperature and precipitation) and host (rodent population density, rodent weight and sex, and larval tick burden) variables were analysed using Generalized Linear Models and multi-model inference to select the best model. Climatic changes (mean annual precipitation and maximum temperature) and individual body mass had a positive effect on hantavirus seroprevalence. Other possible drivers affecting the observed pattern need to be studied further.

This paper argues that most of the problems that actuaries have to deal with in the context of non-life insurance can be usefully cast in the framework of computational intelligence (a.k.a. artificial intelligence), the discipline that studies the design of agents which exhibit intelligent behaviour. Finding an adequate framework for actuarial problems has more than a simply theoretical interest: it also allows a knowledge transfer from the computational intelligence discipline to general insurance, wherever techniques have been developed for problems which are common to both contexts. This has already happened in the past (neural networks, clustering, data mining have all found applications to general insurance) but not systematically, with the result that many useful computational intelligence techniques such as sparsity-based regularisation schemes (a technique for feature selection) are virtually unknown to actuaries.

In this first of two papers, we will explore the role of statistical learning in actuarial modelling. We will show that risk costing, which is at the core of pricing, reserving and capital modelling, can be described as a supervised learning problem. Many activities involved in exploratory analysis, such as data mining or feature construction, can be described as unsupervised learning. A comparison of different computational intelligence methods will be carried out, and practical insurance applications (rating factor selection, IBNER analysis) will also be presented.

This paper examines various forms of individual claim model for the purpose of loss reserving, with emphasis on the prediction error associated with the reserve. Each form of model is calibrated against a single extensive data set, and then used to generate a forecast of loss reserve and an estimate of its prediction error.

The basis of this is a model of the “paids” type, in which the sizes of strictly positive individual finalised claims are expressed in terms of a small number of covariates, most of which are in some way functions of time. Such models can be found in the literature.

The purpose of the current paper is to extend these to individual claim “incurreds” models. These are constructed by the inclusion of case estimates in the model's conditioning information. This form of model is found to involve rather more complexity in its structure.

For the particular data set considered here, this did not yield any direct improvement in prediction error. However, a blending of the paids and incurreds models did so.

Continuing increases in life expectancy beyond previously-held limits have brought to the fore the critical importance of mortality forecasting. Significant developments in mortality forecasting since 1980 are reviewed under three broad approaches: expectation, extrapolation and explanation. Expectation is not generally a good basis for mortality forecasting, as it is subjective; expert expectations are invariably conservative. Explanation is restricted to certain causes of death with known determinants. Decomposition by cause of death poses problems associated with the lack of independence among causes and data difficulties. Most developments have been in extrapolative forecasting, and make use of statistical methods rather than models developed primarily for age-specific graduation. Methods using two-factor models (age-period or age-cohort) have been most successful. The two-factor Lee–Carter method, and, in particular, its variants, have been successful in terms of accuracy, while recent advances have improved the estimation of forecast uncertainty. Regression-based (GLM) methods have been less successful, due to nonlinearities in time. Three-factor methods are more recent; the Lee–Carter age-period-cohort model appears promising. Specialised software has been developed and made available. Research needs include further comparative evaluations of methods in terms of the accuracy of the point forecast and its uncertainty, encompassing a wide range of mortality situations.

The population abundance estimates used in stock assessments or required to establish management measures, depend on the sampling of the entire demographic spectrum of a population resident in a given area. However, for sea turtles, most population estimates are based mainly on nesting beach survey data and only consider a fraction of the population. The Southwest Atlantic Ocean (SWA) is an important foraging and development area for juveniles of the loggerhead sea turtle Caretta caretta where reproductive stocks from various nesting beaches mix. Declines in C. caretta populations have been observed in many parts of the world and bycatch rates of this species in the SWA are among the highest worldwide. This study standardizes the catch rates of loggerheads caught by pelagic longline fisheries in the region, using data collected by observer programs from Brazil and Uruguay. Generalized linear models (GLM) with a delta lognormal approximation were used. The variables used in the model take into account spatial and temporal variations as well as the characteristics of the fleet. In total, 6 272 344 hooks were observed between 1998 and 2007, with minimum effort registered in 2000 (12 010 hooks) and maximum effort in 2005 (1 989 431 hooks). During this period 3778 loggerheads were incidentally captured. The catch rates of loggerheads by the Uruguayan and Brazilian pelagic longline fisheries show oscillations through the years without a clear tendency; however, a low negative trend was observed from 1998 to 2005 with an increase in the last two years (2006 and 2007). The capture per unit of effort (CPUE) values varied between 0.38 to 1.78 ind/1000 hooks in 2005 and 2007, respectively. Distinct zones with differential catch rates were identified, with the higher CPUE values over the continental slope of Uruguay and adjacent waters. The incidental catch rates of this species are influenced, not only by fishing area, but also by year, season, sea surface temperature and gear type. In consequence, these variables and other potential ones should be considered in bycatch estimates by different fisheries because the loggerhead sea turtles are not uniformly distributed. This study intends to contribute not only to the general knowledge of loggerhead sea turtles in the SWA but to a future assessment of their populations at a global scale.

The potential spatial distributions of six epiphytic lichen species were assessed in Switzerland (41 000 km2) as a function of various key climatic drivers and forest types using logistic regression models. Cetrelia cetrarioides is ‘near threatened’, Lobaria pulmonaria is ‘vulnerable’, and Graphis scripta, Hypogymnia physodes, Lecanora cadubriae, Letharia vulpina are not endangered according to the Red List assessment based on IUCN criteria. Lichen presence and absence were derived from the SwissLichens database that contains spatially explicit information on both species presence and absence.

The spatial lichen niches are predicted with R2 values between 0·5 and 0·75 and AUC values between 0·63 and 0·94. Model evaluation shows that the models perform well.

Lichenologists reviewed the spatial predictions of lichen species on the basis of their expert knowledge and concluded that parsimonious regression models may suffice for successful prediction of the potential spatial niche distributions of epiphytic lichen species.

Catch rates (kg hour−1) of horse mackerel (Trachurus trachurus), blue whiting (Micromesistius poutassou) and hake (Merluccius merluccius) from a series of 22 groundfish surveys conducted off Portugal between 1989 and 1999 were analysed using integrated logistic and gamma Generalized Linear Models (GLM). This methodology deals with the large amount of zeros in survey data matrices by modelling the probability of catch and the amount of positive catch separately, and then integrating the two sub-models into a single catch rate model of abundance. Among the explanatory variables included in the models, the geographic areas occupied by fish assemblages, i.e., groups of persistent co-occurring species, explained most of the variability observed for horse mackerel and blue whiting, while depth was the most important factor for hake. Because of hake's ubiquity on the Portuguese margin, models for this species were less parsimonious and explained a lower proportion of total variability compared with the other species. The abundance of horse mackerel is much higher in the winter and reduces with depth. Blue whiting and horse mackerel exhibited opposite trends in abundance from 1991 to 1994, with the latter species presenting a higher abundance throughout this period. No clear temporal trends of abundance were detected for hake. The effect of trawling hour in models suggested that horse mackerel and blue whiting exhibit vertical migrations, staying close to the bottom by day, when abundance is higher. The current approach proved to cope well with sparse catch data matrices and brought out novel information on the ecology of fish species in the Portuguese continental margin by successfully integrating environmental information into the analysis of fish abundance.

In this paper we model the life-history of LTC-patients using a Markovian multi-state model in order to calculate premiums for a given LTC-plan. Instead of estimating the transition intensities in this model we use the approach suggested by Andersen et al. (2003) for a direct estimation of the transition probabilities. Based on the Aalen-Johansen estimator, an almost unbiased estimator for the transition matrix of a Markovian multi-state model, we calculate so-called pseudo-values, known from Jackknife methods. Further, we assume that the relationship between these pseudo-values and the covariates of our data are given by a GLM with the logit as link-function. Since the GLMs do not allow for correlation between successive observations we use instead the “Generalized Estimating Equations” (GEEs) to estimate the parameters of our regression model. The approach is illustrated using a representative sample from a German LTC portfolio.

We consider the problem of claims reserving and estimating run-off triangles. We generalize the gamma cell distributions model which leads to Tweedie's compound Poisson model. Choosing a suitable parametrization, we estimate the parameters of our model within the framework of generalized linear models (see Jørgensen-de Souza [2] and Smyth-Jørgensen [8]). We show that these methods lead to reasonable estimates of the outstanding loss liabilities.