Marginal Changes in Random Parameters Ordered Response Models with Interaction Terms

Marginal changes of interacted variables and interaction terms in random parameters ordered response models are calculated incorrectly in econometric softwares. We derive the correct formulas for calculating these marginal changes. In our empirical example, we observe significant changes not only in the magnitude of the marginal effects but also in their standard errors, suggesting that the incorrect estimation of the marginal effects of these variables as is commonly practiced can render biased inferences on the findings.     

Confidence intervals for impulse responses under departures from normality

Monte Carlo evidence shows that in structural VAR models with fat-tailed or skewed innovations the coverage accuracy of impulse response confidence intervals may deterorate substantially compared to the same model with Gaussian innovations. Empirical evidance suggests that such departures from normality are quite plausible for economic time series. The simulation results suggest that applied researchers are best off using nonparametric bootstrap intervals for impulse responses, regardless of whether or not there is evidence of fat tails or skewness in the error distribution. Allowing for departures from normality is shown to considerably weaken the evidence of the delayed overshooting puzzle in Eichenbaum and Evans (1995).     

An Expert-Based Framework for Probabilistic National Population Projections: The Example of Austria

The traditional way of dealing with uncertainty in population projections through high and low variants is unsatisfactory because it remains unclear what range of uncertainty these alternative paths are assumed to cover. But probabilistic approaches have not yet found their way into official population projections. This paper proposes an expert-based probabilistic approach that seems to meet important criteria for successful application to national and international projections: 1) it provides significant advantages to current practice, 2) it presents an evolution of current practice rather than a discontinuity, 3) it is scientifically sound, and 4) it is applicable to all countries.In a recent Nature article (Lutz et al., 1997) this method was applied to 13 world regions. This paper discusses the applicability to national projections by directly taking the alternative assumptions defined by the Austrian Statistical Office. Sensitivity analyses that resolve some methodological questions about the approach are also presented.     

The state intervenes in the battle of the sexes: Causal effects of paternity leave

Do family policies influence attitudes and behavior or are they merely reflections of pre-existing attitudes? We consider the implementation of the Norwegian daddy quota, 4 weeks of parental leave reserved for the father, as a natural experiment, and examine the long-run causal effects on attitudes toward gender equality, on conflicts and sharing of household labor, and on support for public childcare. We find that respondents who had their last born child just after the reform report an 11% lower level of conflicts over household division of labor and that they are 50% more likely to equally divide the task of washing clothes than respondents who had their last child just before the reform.     

Comparing relative effects of education and economic resources on infant mortality in developing countries

Research on the social determinants of health has often considered education and economic resources as separate indicators of socioeconomic status. From a policy perspective, however, it is important to understand the relative strength of the effect of these social factors on health outcomes, particularly in developing countries. It is also important to examine not only the impact of education and economic resources of individuals, but also whether community and country levels of these factors affect health outcomes. This analysis uses multilevel regression models to assess the relative effects of education and economic resources on infant mortality at the family, community, and country level using data from demographic and Health Surveys in 43 low-and lower-middle-income countries. We find strong effects for both per capita gross national income and completed secondary education at the country level, but a greater impact of education within families and communities.     

When what you have is not enough—Acquiring Australian qualifications to overcome non-recognition of overseas skills

Skilled migration is an important strategy in developed economies seeking to address skills shortages and population ageing. Research on the labour market outcomes of skilled migrants tends to focus on employers' devaluation of skills without considering the role of immigration policy in the migration process. Moreover, there is little understanding of whether efforts to meet employer demands for local qualifications improve labour market outcomes. Drawing on a study on skilled migrants sponsored under the State-Specific and Regional Migration Scheme in the regional state of South Australia, we explore the shaping of skills and skills recognition in the migration journey, particularly migrants' strategy of reskilling in response to employer demands for local qualifications. Our logistic regressions on the association between the acquisition of Australian qualifications and labour market outcomes reveal only marginal returns to these efforts. We argue that Australia should consider developing a more coherent skilled migration process to better harness the human capital of skilled migrants.     

The geometry of Gaussian double Markovian distributions

Gaussian double Markovian models consist of covariance matrices constrained by a pair of graphs specifying zeros simultaneously in the matrix and its inverse. We study the semi-algebraic geometry of these models, in particular their dimension, smoothness, and connectedness as well as algebraic and combinatorial properties.     

Sensitivity Analysis Using Risk Measures

In a quantitative model with uncertain inputs, the uncertainty of the output can be summarized by a risk measure. We propose a sensitivity analysis method based on derivatives of the output risk measure, in the direction of model inputs. This produces a global sensitivity measure, explicitly linking sensitivity and uncertainty analyses. We focus on the case of distortion risk measures, defined as weighted averages of output percentiles, and prove a representation of the sensitivity measure that can be evaluated on a Monte Carlo sample, as a weighted average of gradients over the input space. When the analytical model is unknown or hard to work with, nonparametric techniques are used for gradient estimation. This process is demonstrated through the example of a nonlinear insurance loss model. Furthermore, the proposed framework is extended in order to measure sensitivity to constant model parameters, uncertain statistical parameters, and random factors driving dependence between model inputs.     

Variable selection for generalized linear mixed models by L 1-penalized estimation

Generalized linear mixed models are a widely used tool for modeling longitudinal data. However, their use is typically restricted to few covariates, because the presence of many predictors yields unstable estimates. The presented approach to the fitting of generalized linear mixed models includes an L ₁-penalty term that enforces variable selection and shrinkage simultaneously. A gradient ascent algorithm is proposed that allows to maximize the penalized log-likelihood yielding models with reduced complexity. In contrast to common procedures it can be used in high-dimensional settings where a large number of potentially influential explanatory variables is available. The method is investigated in simulation studies and illustrated by use of real data sets.     

Insurance demand and social comparison: An experimental analysis

This paper analyzes whether social comparison can explain the low take-up of disaster insurance usually reported in field studies. We argue that risks in the case of disasters are highly correlated between subjects whereas risks for which high insurance take-up can be observed (e.g. extended warranties or cell phone insurance) are typically idiosyncratic. We set up a simple model with social reference points and show that in the presence of inequality aversion social comparison makes insurance indeed less attractive if risks are correlated. In addition we conducted a simple experiment which confirms these theoretical results. The average willingness to pay for insurance is significantly higher for idiosyncratic than for correlated risks.