فصلنامه پژوهشهای اقتصادی ایران (Mar 2024)
Transmission Mechanisms of Contemporaneous Risk in Investment Portfolios: An R2 Connectedness Approach With Evidence From the Iranian National Pension Fund Investment Company
Abstract
One of the primary concerns of the Iranian National Pension Fund is managing its investment portfolio. In this respect, the present study aimed to examine the long-term investment portfolio, the largest subset of which is V-sandoq. The analysis used the R2 connectedness approach proposed by Naeem et al. (2023) over the period from September 17, 2013, to September 22, 2023. The study focused on the immediate influence and susceptibility to influence of the stocks within the National Pension Fund. The results showed that, in terms of net influence and susceptibility, the stocks of Group 1 (i.e., Kechad, Foulad, Kegol, and Sheranol) were the most influential, transferring risk to the network. Conversely, the stocks of Group 2 (i.e., Shepas, Pasa, Shekabir, and Vebshahr) were the most influenced by the network. Therefore, risk is transferred from Group 1 stocks to the network, impacting Group 2 stocks the most. In network analysis, during a bear market with a threshold of -4%, there is a high degree of connectivity among the stocks in the portfolio. This suggests that portfolio adjustments are necessary under bear market conditions. Conversely, in a bull market with a threshold of +4%, there is no connectivity between the stocks, indicating that no portfolio adjustments are needed under such conditions.1.IntroductionIn recent years, Iran has consistently faced challenges with pension funds and the inability to generate adequate income to pay retirement salaries. With the number of retirees expected to increase in the coming years (particularly from the 1980s generation), effective management of the investment portfolio of the National Pension Fund’s subsidiaries has become increasingly critical. Many state-owned companies were transferred to the National Pension Fund to finance retired pay from their profitability. However, budget evidence indicates that over 80% of retirement salaries are still financed through the government budget. This underscores the importance and necessity of revising the investment portfolio of the National Pension Fund’s investment holdings. In this respect, the present study aimed to examine the portfolio management of one of the largest subsidiaries of the National Pension Fund, namely the Investment Company of the National Pension Fund or V-sandoq, over the period from September 17, 2013, to September 22, 2023. The study used the vector autoregression (VAR) model with time-varying parameters and R2 connectedness, as an immediate response, proposed by Naeem et al. (2023). The immediate impact analysis of variables on/from each other was chosen because any national, regional, or global event has immediate effects, and providing an appropriate response in portfolio management is of great importance.2.Materials and MethodsThe study employed the TVP-VAR algorithm and the Kalman filter introduced by Antonakakis et al. (2020), in conjunction with the approach proposed by Naeem et al. (2023). The key econometric structure of the TVP-VAR model is outlined below. For the sake of simplicity, it is presented in the form of a first-order VAR. Thus, the TVP-VAR model is as follows:(1) (2) Time-varying parameters and time-varying error variances are essential components for the generalized impulse response functions (GIRF) and generalized forecast error variance decomposition (GFEVD) developed by Koop et al. (1996) and Pesaran and Shin (1998). These components underpin the connectivity approach of Diebold and Yılmaz (2012, 2014). To obtain GIRF and GFEVD, the TVP-VAR needs to be converted to TVP-VMA by applying the Wold representation theorem. According to this theorem, GIRFs i j,t (K) at a forecast horizon K do not assume or depend on the ordering of shocks, providing a more robust interpretation of VAR models compared to standard IRFs, which are sensitive to the order of variables in the econometric system. The GIRF approach reflects the dynamic differences between all variables jjj. Mathematically, it can be expressed as Equation (3):(3) (4) Subsequently, GFEVD ψij,t(K)\psi_{ij,t} (K)ψij,t(K) represents the unique contribution of each variable to the forecast error variance of variable iii, interpreted as the percentage impact of one variable on the forecast error variance of another variable. This can be expressed as Equation (5):(5) The criteria for GIRF and GFEVD can help determine how much variable iii is influenced by others and how much it influences others. Three metrics are used for this purpose.First, we must determine how much other variables in the system influence variable iii. This is obtained by summing the error variance shares for variable iii relative to variable jjj. The influence from others is then calculated using Equation (6):(6) Second, the impact of variable iii on others in the system is calculated through the measurement known as influence on others. This measurement is derived by summing the effects (error variance) that variable iii imposes on the forecast error variance of other variables:(7) The total connectivity index (TCI) is calculated based on the Monte Carlo simulations presented by Chatzanzinou et al. (2021). It demonstrates that the self-variance share consistently exceeds or equals all cross-variance shares. Since the average co-movement of the network is expressed as a percentage, which should be between [0,1], TCI needs to be slightly adjusted:(8) Finally, the TCI definition is modified to obtain pairwise partial connectivity index (PCI) scores between variables iii and jjj as follows:(9) 3.Results and DiscussionFigure 2 illustrates the temporal dynamics of stock influences received from other stocks. It shows the extent to which each stock has transferred or received risk from others. The stocks above the zero line indicate a net influence on the network, while those below indicate a net reception from the network during the examined period. Notably, Kechad, Foulad, Kegol, and Sheranol (Group 1) predominantly acted as influencers, transferring risk to the network. In contrast, Shepas, Pasa, Shekabir, and Vabshahr (Group 2) exhibited the highest reception from the network. Therefore, it can be inferred that external shocks transfer risk from Group 1 to the network, notably impacting the stocks in Group 2.It is crucial to recognize that this influence/reception patterns vary over time and exhibit significant fluctuations. Specifically, the chart shows that the influence/reception of stocks on/from the network decreased with the outbreak of the COVID–19 pandemic from January 19, 2021. Conversely, the disclosure of the letter regarding the increase in petrochemical feed rates on May 7, 2023 heightened the risk transfer from petrochemical stocks to the studied network. This underscores that external shocks do not uniformly affect the portfolio under review, necessitating separate examination of each. Figure1: Net Influence/Reception of Stocks on/from Each Other Source: Research findings4.ConclusionThe results of the long-term portfolio analysis indicated varying levels of interconnectedness influenced by economic, political, military, and health conditions—with the connectivity averaging around 45%. This reflects a high risk for the long-term portfolio. In terms of net influence and reception, Kechad, Foulad, Kegol, and Sheranol (Group 1) generally exerted influence by transferring risk to the network. In contrast, Shepas, Pasa, Shekabir, and Vabshahr (Group 2) predominantly received risk from the network. Thus, during external shocks, risk tends to shift away from Group 1 stocks, thus impacting Group 2 significantly. The outbreak of the COVID–19 pandemic on January 19, 2021 led to a decrease in the influence/reception of stocks on or from the network. Conversely, the disclosure of an increase in petrochemical feed rates on May 7, 2023 heightened risk transfer from petrochemical stocks to the studied network. Concerning the network analysis, there is a high degree of connectivity among the stocks in the portfolio during a bear market with a threshold of -4%. This suggests that portfolio adjustments are necessary under bear market conditions. In bearish markets, it thus becomes imperative to select stocks that have less connectivity. On the contrary, in a bull market with a threshold of +4%, there is no connectivity between the stocks, indicating that no portfolio adjustments are needed under such conditions. Hence, while the examined portfolio is optimal during bull markets, adjustments are essential during bear markets to mitigate risks associated with high connectivity.
Keywords
