A new variable importance measure for random forests with missing data

Random forests are widely used in many research fields for prediction and interpretation purposes. Their popularity is rooted in several appealing characteristics, such as their ability to deal with high dimensional data, complex interactions and correlations between variables. Another important feature is that random forests provide variable importance measures that can be used to identify the most important predictor variables. Though there are alternatives like complete case analysis and imputation, existing methods for the computation of such measures cannot be applied straightforward when the data contains missing values. This paper presents a solution to this pitfall by introducing a new variable importance measure that is applicable to any kind of data—whether it does or does not contain missing values. An extensive simulation study shows that the new measure meets sensible requirements and shows good variable ranking properties. An application to two real data sets also indicates that the new approach may provide a more sensible variable ranking than the widespread complete case analysis. It takes the occurrence of missing values into account which makes results also differ from those obtained under multiple imputation.     

A toolbox of permutation tests for structural change

The sup $LM$ test for structural change is embedded into a permutation test framework for a simple location model. The resulting conditional permutation distribution is compared to the usual (unconditional) asymptotic distribution, showing that the power of the test can be clearly improved in small samples. Furthermore, the permutation test is embedded into a general framework that encompasses tools for binary and multivariate dependent variables as well as model-based permutation testing for structural change. It is also demonstrated that the methods can not only be employed for analyzing structural changes in time series data but also for recursive partitioning of cross-section data. The procedures suggested are illustrated using both artificial data and empirical applications (number of youth homicides, employment discrimination data, carbon flux in tropical forests, stock returns, and demand for economics journals).     

Trade Union Preferences in Double Dividend Models

This paper analyses wage formation in a unionized economy where consumers as an externality in their utility function have the level of local pollution. If modelled in a microeconomically consistent way this externality should also be present in the preferences of the trade union. The key result is that when this trade‐off between pollution and employment is included in the trade unions’ preferences then they are willing to lower wages to generate substitution towards higher employment and lower pollution. As a consequence, an increase in the pollution tax will lead to lower wages. At a more general level the results show that in models analyzing pollution issues, such as the double dividend literature, it is very important for the policy conclusions how trade unions are introduced.     

ADJUSTMENT TO PRODUCTION UNCERTAINTY AND THE THEORY OF THE FIRM: A COMMENT

This paper reconsiders the model presented by Flacco and Kroetch [1986]. It shows that with additive technological uncertainty the firm will produce and contract to sell the same output as under certainty. When the Flacco/Kroetch model is generalized to allow the firm to select contract commitments and planned output ex ante, no simple comparisons with certainty are possible. These results are broadly consistent with those obtained by Mills [1959; 1962] for demand uncertainty.     

When should there be which kind of technology assessment? A plea for a strictly problem-oriented approach from the very outset

Technology assessment (TA) is generally classified as problem-oriented and thus transdisciplinary research. This is due to the fact that the aim of TA is to work out solutions for problems outside science in order to offer advice to its addressees, namely those working in politics and science and members of society in general. In this paper, we propose that the problem-oriented approach also be used as the basis for the decision regarding when a TA should be conducted in a particular situation, and which TA should be used. As a first step to this end, a broad range—almost like a kind of coarse radar—and large number of topics or problem sketches are created, which then serve as a basis for a discursive process of negotiation, during which topics are to be carried over into a detailed problem analysis (“fine radar”). These detailed problem analyses, which are important for transdisciplinary research, represent in themselves small research projects (pre-projects). They provide, in addition, some initial indications for a methodologically promising project design. The methodological approach described here is currently being tested in a procedure in which TA topics are being monitored for the German Federal Ministry of Education and Research (Bundesministerium für Bildung und Forschung, BMBF).     

Demographic change and the labour share of income

Despite similar levels of per capita income, education and technology, the development of labour income shares in OECD countries has displayed different patterns since 1960. The paper examines the role of demography in this regard. We first use a standard overlapping generations model to derive the mechanisms by which demographic change can affect the labour share. It turns out that demographic change can affect the labour share either by altering the domestic capital intensity, by causing factor-biased technological change or in a small open economy framework by creating a gap between domestic savings and investments. The latter affects the country’s investments abroad and in return its net foreign asset income which directly leads to changes in the labour share. Empirical estimations based on these insights, provide evidence that an increases in the expected retirement durations and old-age dependency ratios as well as declines in labour force growth rates have indeed been major forces behind the decline in labour shares that took place in many countries. These effects tend to be larger in open economies and pension reforms towards a funded pension system seem to have accelerated the effects.     

Management Support for Portfolio Companies of Venture Capital Firms: An Empirical Study of German Venture Capital Investments

To date, there is a dearth of research on strategic antecedents and consequences of management support activities which German venture capital firms (VCFs) provide for the portfolio companies (PCs) they have invested in. This article provides insights regarding such support practices, their main determinants and impacts on the economic performance of PCs. Hypotheses were derived regarding usage of various types of management support, typical functional foci of management support and the influence of VCF and PC characteristics on management support practice. Hypotheses also cover relationships between type, intensity, and frequency of management support and PC performance. Data for 103 PCs of 12 German VCFs indicate the VCFs can improve the performance of their PCs by providing consultative management support, including active involvement in key functional decisions.     

A simple theory of ‘meso’. On the co-evolution of institutions and platform size—With an application to varieties of capitalism and ‘medium-sized’ countries

This paper deals with institutional emergence in the well-known ‘evolution of cooperation’ framework and focuses on its size dimension. It is argued that some ‘meso’ (rather than ‘macro’) level (to be numerically determined) is the proper level of cultural emergence, diffusion, and retention. Also Schumpeterian economists (K. Dopfer et al.) have discussed institutions as ‘meso’ phenomena, and Schelling, Axelrod, Arthur, Lindgren, and many others have dealt with ‘critical masses’ of coordinated agents, including related segregations of populations. However, the process and logic of emergent group size has rarely been explicitly explored so far. In this paper, ‘meso’ will be explained, in an evolutionary and game–theoretic frame and a population perspective, in terms of a cooperating group smaller than the whole population involved. Mechanisms such as memory, monitoring, reputation chains, and active partner selection will loosen the total connectivity of the deterministic ‘single-shot’ benchmark and thus allow for emergent ‘meso’-sized arenas, while expectations to meet a cooperative partner next round remain sufficiently high. Applications of ‘meso-nomia’ include the deep structure of ‘general trust’ and the surprisingly high macro-economic and macro-social performance in ‘small’ and ‘well-networked’ countries which helps to explain persistent ‘varieties of capitalism’. A strategy for empirical application of the theoretical approach and some first empirical indications of its relevance are presented.