Публикации

The Minkowski weighted K-means (MWK-means) is a recently developed clustering algorithm capable of computing feature weights. The cluster-specific weights in MWK-means follow the intuitive idea that a feature with low variance should have a greater weight than a feature with high variance. The final clustering found by this algorithm depends on the selection of the Minkowski distance exponent. This paper explores the possibility of using the central Minkowski partition in the ensemble of all Minkowski partitions for selecting an optimal value of the Minkowski exponent. The central Minkowski partition appears to be also a good consensus partition. Furthermore, we discovered some striking correlation results between the Minkowski profile, defined as a mapping of the Minkowski exponent values into the average similarity values of the optimal Minkowski partitions, and the Adjusted Rand Index vectors resulting from the comparison of the obtained partitions to the ground truth. Our findings were confirmed by a series of computational experiments involving synthetic Gaussian clusters and real-world data

We examine 597 estimates of habit formation reported in 81 published studies. The mean reported strength of habit formation equals 0.4, but the estimates vary widely both within and across studies. We use Bayesian and frequentist model averaging to assign a pattern to this variance while taking into account model uncertainty. Studies employing macro data report consistently larger estimates than micro studies: 0.6 vs. 0.1 on average. The difference remains 0.5 when we control for 30 factors that reflect the context in which researchers obtain their estimates, such as data frequency, geographical coverage, variable definition, estimation approach, and publication characteristics. We also find that evidence for habits strengthens when researchers use lower data frequencies, employ log-linear approximation of the Euler equation, and utilize open-economy DSGE models. Moreover, estimates of habits differ systematically across countries.

We analyse the profession of criminal defence lawyers (“advocates”) in Russia to understand their potential for collective action in an imperfect institutional environment. In 2013, we conducted a survey of 372 advocates in 9 regions of Russia. The following two main hypotheses are tested: (1) lawyers with strong ethical values have a higher demand for collective action; and (2) the negative experience of clients' rights violations by law enforcement officers can motivate advocates to support the foundation of a strong professional association. We suggest that an advocate's profession with bona fide members at the core could be an instrument to evaluate and to improve the quality of law enforcement in Russia.

Is in utero exposure to testosterone correlated with earnings? The question matters for understanding determinants of wage differences that have attracted so much attention among economists in the past decade. Evidence indicates that markers for early testosterone exposure are correlated with traits like risk-taking and aggressiveness. But it is not at all clear how such findings might map into labor market success. We combine unique data from the Russian Longitudinal Monitoring Survey with measured markers (2D:4D ratios) for testosterone exposure and find that lower digit ratios (higher T) correlate with higher wages for women and for men, when controlling for age, education and occupation. There is also some evidence of a potential non-linear, inverse U-effect of digit ratios on wages but this is sensitive to choice of specification. These findings are consistent with earlier work on prenatal T and success in careers (Coates et al., 2009) but inconsistent with the work of Gielen et al. (2016) who find differing effects for men and women.