© 2016 Nicolaus Copernicus University. All rights reserved. http://www.dem.umk.pl/dem D Y N A M I C E C O N O M E T R I C M O D E L S DOI: http://dx.doi.org/10.12775/DEM.2016.007 Vol. 16 (2016) 117−131 Submitted November 30, 2016 ISSN (online) 2450-7067 Accepted December 20, 2016 ISSN (print) 1234-3862 Krzysztof Kompa, Dorota Witkowska * Performance of Pension Funds and Stable Growth Open Investment Funds During the Changes in the Polish Retirement System A b s t r a c t. The conditions of the pension funds (OFE) functioning were essentially changed in the years 2011–2014. The aim of the paper is to find out if these modifications influence the efficiency of the pension funds and to compare the performance of these funds to stable growth open investment funds (FIO). The analysis is provided for selected funds in the years 2009–2015. We conclude that in the examined period, OFE performed better than FIO, and the modifications of the rules for the pension funds caused the increase of risk and decrease of investment efficiency of these funds’ portfolios. K e y w o r d s: pension funds, stable growth open investment funds, investment efficiency, Sharpe model, CAPM, Sharpe, Treynor and Jensen ratios. J E L Classification: G11; C12. Introduction The pension system in communist Poland was the defined benefit and pay-as-you-go (PAYG) scheme. However, the demographic changes, pen- sion privileges concerning more and more occupational groups and econom- ic sectors, together with the so-called early retirement regulations, which * Correspondence to: Krzysztof Kompa, Warsaw University of Life Sciences – SGGW, Department of Econometrics and Statistics, 166 Nowoursynowska Street, 02-787 Warsaw, Poland, e-mail: Krzysztof_kompa@sggw.pl; Dorota Witkowska, University of Lodz, De- partment of Finance and Strategic Management, 22/26 Matejki Street, 90-237 Lodz, Poland, Dorota.witkowska@uni.lodz.pl Krzysztof Kompa, Dorota Witkowska DYNAMIC ECONOMETRIC MODELS 16 (2016) 117–131 118 went into force in late 70-ties 1 , caused the increase of the pension system costs. In 1981 the pension contribution was 25% of wages, in the years 1987–1989 it rose to 38%, and obtained the level of 45% in 1990 (see (Wojciechowski 2011, Podstawka 2005, p. 259)). The increasing deficit in the Polish pension system enforced its trans- formation, which took place in 1999. The new pension system replaced de- fined benefit scheme by defined contribution one, it enriched the PAYG system by the mandatory capital-funded pillar, and introduced voluntary plans. New regulations were also to abolish sectoral and occupational privi- leges and early retirement programs. The reformed retirement system has been consisted in three pillars – Social Insurance Institution (ZUS), open pension funds (OFE) and voluntary capital-funded system. The pension con- tribution (i.e. 19.52% of earnings) was divided between both mandatory pillars – ZUS 12.22% and OFE 7.3%. The first essential manipulation in the original pension reform was made in 2011 when the contribution to pension funds was diminished from 7.3% to 2.3%. The second and the most drastic regulation, consisted in shifting 51.5% of the assets, held by the pension funds, to the Social Insurance Insti- tution (including all debt securities issued and guaranteed by the State Treasury). Overhaul of the pension system also concerned changes in the OFEs’ investment portfolio since private pension funds have no longer been allowed to invest in government bonds. These new law (from 2013) went into effect in February 2014. The third significant modification took place in 2014 and changed the character of pension funds which have been no longer obligatory. According to the new regulations, each employee has had four months every four years to decide whether 2.92 percent of their income goes to a chosen private fund or to ZUS. It means that after all mentioned above regulations the part of the pension contribution transferred to the pension funds decreased from 37.4% to 15% or even zero. There have been numerous studies concerning the efficiency of mutual and pension funds operating in Poland. Mutual funds’ performance is ana- lyzed by Karpio and Żebrowska-Suchodolska (2010, 2011, 2012), Ostrowska (2003), Perez (2012), Witkowska (2009), Witkowska et al. (2009) and Zamojska (2012). Whereas evaluation of pension funds invest- ment activity is made by Białek (2009), Chybalski (2006, 2009), Dybał (2008), Karpio and Żebrowska-Suchodolska (2014, 2016), Kompa and 1 Early retirement regulations decreased the real pension age by several years since em- ployees who fulfil certain conditions were allowed to obtain pension benefits earlier i.e. be- fore statutory retirement age. Performance of Pension Funds and Stable Growth Open Investment Funds… DYNAMIC ECONOMETRIC MODELS 16 (2016) 117–131 119 Wiśniewski (2015), Kompa and Witkowska (2015, 2016), Marcinkiewicz (2009) and Witkowska, Kompa (2012). The mentioned research was provid- ed for differently defined time spans, length of the samples, various frequen- cy of measurement, taking into account bear and bull markets, and variety of efficiency measures. However, the investigation concerning the influence of political decisions to the performance of pension and investment funds mar- ket is rare. There is also lack of profound comparable research of the effi- ciency of pension and mutual funds. This paper is to fulfill that gap in literature. The aim of our research is to find out if mentioned above changes of the OFEs’ functioning influence the efficiency of the pension funds and to compare the performance of these funds to stable growth investment funds (FIO). The analysis is provided for selected funds applying statistical inference, Sharpe and capital assets pric- ing (CAPM) models and classical investment efficiency ratios. 1. Data and Methods To answer the question about the consequences of manipulations in the pen- sion system, the whole period of analysis, denoted by A, is divided into three pairs of sub-periods according to the moments when new regulations went into effect:  B, decreasing of the contribution transferred to OFE,  C, shifting of assets from OFE to ZUS and changes in the OFEs’ portfo- lio composition,  D, waving the obligation of pension funds membership. Empirical analysis of the investment efficiency is provided on the basis of daily observations from the years 2009–2015. The considered time span covers seven samples:  A from 1.01.2009 to 31.12.2015 (84 months),  B1 from 1.01.2009 to 30.04.2011, and B2 from 1.05.2011 to 31.08.2013 (28 months each sub-period),  C1 from 1.04.2012 to 31.01.2014, and C2 from 1.02.2014 to 31.12.2015 (22 and 23 months respectively),  D1 from 1.01.2013 to 30.06.2014, and D2 from 1.07.2014 to 31.12.2015 (18 months each). In our research we use daily logarithmic rate of returns evaluated from:  participation units of stable growth open investment funds and account- ing units of pension funds, which are managed by the same six invest- ment and pension funds companies, namely: Allianz, Aviva, Nationale Krzysztof Kompa, Dorota Witkowska DYNAMIC ECONOMETRIC MODELS 16 (2016) 117–131 120 Nederlanden, PEKAO, PKO and PZU Investment and Pension Funds Companies 2 ,  Warsaw Stock Exchange Index – WIG,  Poland’s Official Treasury Bonds Index – TBSP.Index, and  Warsaw Interbank Offered Rate – WIBOR. WIG, TBSP and WIBOR (being the interest rate of 3 months loans) are used as benchmarks 3 in our analysis since they represent capital, treasury bond and money markets, respectively. WIG is treated as the market index whereas TBSP and WIBOR – as risk free instruments. Analysis of returns and risk generated by the investment portfolios con- structed by selected funds is conducted using statistical interference (assum- ing the significance level 0.05). We will verify 4 the following null hypothe- ses concerning: rates of return risk Sharpe model / CAPM E(ROFE) = 0; E(RFIO) = 0; E(Rbenchmark) = 0  = 0;  = 0 E(ROFE) = E(RFIO) D2(ROFE) = D2(RFIO) OFE = FIO E(Rbefore) = E(Rafter) D2(Rbefore) = D2(Rafter) before = after where, E(R) – expected returns, D 2 (R) – variance of returns, ROFE, RFIO – returns from OFE and FIO respectively, ,  – parameters of Sharpe model or CAPM, Rbefore, Rafter, before, after – returns from the portfolio and beta coefficients before and after the change went into effect, respectively. The comparison of the funds’ efficiency is provided applying Sharpe, Treynor and Jensen ratios which are evaluated for all considered pension and mutual funds in all analyzed time spans, and for differently defined risk free instruments. 2. Rates of Return In the first step of our analysis we check out if daily rates of return are significantly positive or negative. It is visible (Table 1) that expected returns from all considered pension funds’ investment are significantly positive in 2 The selection of mentioned above Investment and Pension Funds Companies is connect- ed with the investigations previously provided by the authors separately for pension funds and investment funds. 3 It is worth mentioning that WIG and WIBOR are used to calculate benchmarks, which are used to evaluate pension funds efficiency (Dziennik Ustaw Rzeczypospolitej Polskiej 2014, poz. 753). 4 Description of all applied tests is presented in (Tarczyński, Witkowska and Kompa, 2013, p. 18–19, 25, 72) Performance of Pension Funds and Stable Growth Open Investment Funds… DYNAMIC ECONOMETRIC MODELS 16 (2016) 117–131 121 the whole period of consideration (denoted by A), and for the periods B1 and C1, describing performance of pension funds before the essential changes of their functioning. It is also noticeable that in the periods C2 and D2, i.e. after the modifications went into effect, the average rates of return are negative, although the null hypothesis cannot be rejected for the significance level 0.05. Table 1. Test statistics verifying the hypothesis about expected returns from pension funds in considered periods H0: E(ROFE) = 0 Period A B1 B2 C1 C2 D1 D2 Number of observations 1823 604 610 480 499 390 393 Allianz 1.9306 2.7360 1.3129 2.4911 –0.1800 0.7590 –0.3116 Aviva 1.7404 2.6509 1.2352 2.2974 –0.2728 0.7827 –0.4601 Nationale Nederlanden 1.7662 2.6364 1.3028 2.4438 –0.3501 0.8735 –0.5018 PEKAO 1.5230 2.7997 1.0987 2.2557 –0.4316 0.7572 –0.6005 PKO 2.0540 2.9102 1.4263 2.6620 –0.1030 0.9630 –0.2776 PZU 1.7043 2.5413 1.0815 2.2353 –0.1397 0.8623 –0.3361 Note: Bold letters denote rejection of null hypothesis. Table 2. Test statistics verifying the hypothesis about expected returns from stable growth investment funds in considered periods H0: E(RFIO) = 0 Period A B1 B2 C1 C2 D1 D2 Number of observations 1823 604 610 480 499 390 393 Allianz 0.0849 0.5640 –0.1806 0.1617 –0.5957 –0.2008 –0.8648 Aviva 2.0625 1.9847 0.7763 1.8784 0.0660 1.0115 –0.0550 Nationale Nederlanden 1.7755 1.8450 0.6528 1.6648 0.2879 0.6511 –0.2487 PEKAO 0.3032 1.6871 –1.0786 0.6526 –0.3112 0.1099 –0.5922 PKO 2.1753 1.8920 0.8929 1.8669 0.5899 0.9977 0.1458 PZU 1.1236 1.5555 0.3250 1.3278 –0.6382 0.5266 –0.6100 Note: Bold letters denote rejection of null hypothesis. Stable growth investment funds’ performance (Table 2) seems to be worse than the pension funds since only Aviva, Nationale Nederlanden and PKO generated significantly positive returns in the periods A, B1 and C1. Howev- er, the comparison of returns, using Cochran-Cox test (Table 3), shows that differences between returns obtained by both types of funds are not signifi- cant, except PEKAO in B2 when OFE performed better than mutual fund. Analyzing returns from the benchmarks (Table 4) one may notice that interest rate WIBOR generated positive returns in all periods, bond market performed well with significantly positive returns in all periods but D1 and D2, whereas expected returns from WIG do not significantly differ from zero, except the period B1. Krzysztof Kompa, Dorota Witkowska DYNAMIC ECONOMETRIC MODELS 16 (2016) 117–131 122 Table 3. Test statistics for comparison of expected returns H0: E(ROFE) = E(RFIO) and risk H0: D 2 (ROFE) = D 2 (RFIO) for pension and investment funds Period Hypotheses Allianz Aviva NN PEKAO PKO PZU A Return 1.1902 –0.0116 0.1989 1.0688 0.2125 0.0000 Variance 1.0935 1.1017 1.1702 1.0740 1.3329 1.5200 B1 Return 0.8956 –0.0783 0.1499 0.1975 0.5473 –0.2761 Variance 1.6486 1.4068 1.2048 1.4157 1.0283 1.9268 B2 Return 0.8969 0.3501 0.3934 1.6668 0.3259 0.2907 Variance 1.4784 1.0615 1.0968 1.3140 1.0511 2.0000 C1 Return 1.9833 0.2024 0.6182 0.8171 0.7680 –0.2969 Variance 1.1808 1.0581 1.2008 1.1005 1.2324 2.0000 C2 Return 0.2589 –0.3431 –0.4286 –0.1533 –0.2260 0.4545 Variance 2.0667 2.1169 2.1375 2.0351 2.4110 0.9857 D1 Return 0.8041 0.0796 0.2887 0.5390 0.2098 0.0000 Variance 1.6271 1.3958 1.4758 1.3195 1.7198 1.5306 D2 Return 0.0885 –0.2676 –0.4538 –0.2788 –0.3561 0.2107 Variance 2.0134 2.1991 2.1157 2.0944 2.3616 1.0462 Note: NN is an abbreviation of Nationale Nederlanden. Bold letters denote rejection of null hypothesis. Shadowed cells denote situation when D 2 (ROFE)