1 / 44

Some stylized facts of Russian private pension funds

Some stylized facts of Russian private pension funds. Didenko Alexander International Financial Laboratory Alexander.didenko@gmail.com. Questions. W hat funds are efficient? What metrics to use? Is there any persistence? Do they inform customers about risks?

york
Télécharger la présentation

Some stylized facts of Russian private pension funds

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Some stylized facts of Russian private pension funds Didenko Alexander International Financial Laboratory Alexander.didenko@gmail.com

  2. Questions • What funds are efficient? • What metrics to use? • Is there any persistence? • Do they inform customers about risks? • Do they have behavioral biases?

  3. Dataset and methods • 30 quarters * 30 private pension funds • IIIQ’ 05 – IVQ’ 12 • Data Envelopment Analysis • Malmquist productivity index • T- and KS-tests • Granger causality

  4. DEA - conceptual model Input 1 Output 1 Production Plans Input 2 Output 1 Input N Output 1

  5. Data envelopment analysis • We have j DMUs • Which use v inputs x • To produce u outputs y • DEA-efficiency is defined as a ratio of a weighted sum of outputs to a weighted sum of inputs

  6. Example from Coopers et al.

  7. Malmquist index • Decomposition of dynamic DEA to three components: • technical efficiency change on the best practice technologies • change in scale efficiency • technical change measured as a shift in the benchmark technology • which sum to total change

  8. DEA – general model for funds Financial Capital Return Pension Funds Risk Market Share Human Capital

  9. DEA – our specificaion Active return CVaR Pension Funds NAV Share E+R Ratio Diversification

  10. CVaR • Wuertz, Chalabi, Chen, Ellis (2009); • RUPAI, RUPCI, RGBI • Alpha=0.05 • Weekly data • Average quarterly CVaR

  11. Diversification • There are plenty of D. measures • We use that of Goetzmann, Kumar, 2008

  12. H1. Funds convey useful info in names • “professionally-looking” terms to indicate attitude to risk • “Balanced” • “Aggressive”, • etc. • do funds really inform potential contributors about riskiness? • we classified funds by 5 categories of riskiness based on names • affinity between CVaRs distribution of 5 classes • affinity of random subsamples inside classes • two-sample Kolmogorov-Smirnov and Student’s t tests

  13. Affinity of CVaR distributions • Classes 1, 2, 3 are way more homogeneous than any other class or total sample • Classes 1 and 3 are very close • Class 4 is similar to class 2 and class 3 • Only class 5 is REALLY different: • Distinctive both by T and KS measures • Homogeneous (after many resamplings)

  14. H1. Busted/plausible? BUSTED!

  15. H2. Are funds prone to herding? • We have information about aggregated portfolio structure • We can test for • Correlation • Granger causation • in changes of portfolio shares • Between funds and between quartiles of capitalization/efficiency

  16. Granger causality: equities

  17. Sum of causation in eq.chng by fund

  18. Sum of causation by cap quartile

  19. We tested the same for: • Malmquist efficiency quartiles • All 4 submeasures • No result • Matrix of granger causation for randomly generated matrices with same proportions, means, sd’s • Results are similar to real granger-causation matrices

  20. H2. Herding/!Herding? PLAUSIBLE

  21. What specification to use? • DRS, VRS, IRS, CRS, FDH? • Input/output/two-way? • We want to have some predictable measure • to have good logit-regression, we need sample with some funds efficient and some – not • too much “efficiency” => bad

  22. Dea

  23. Malmquist productivity • Same questions about specification • For our results be comparable • we have to use the same set of specifications for DEA and Malmquist productivity

  24. Window dressing? Wow!

  25. Dropping expense+reward ratio

  26. H2. Funds do not window-dress? PLAUSIBLE

More Related