“Forecasting the net investment position based on conventional and ESG stock market indices: The case of Ukraine and Austria”

This paper examines the relationship between traditional and ESG stock market indices and the net international investment position for the case of Austria and Ukraine. For these purposes, the following methods are used: variance analysis, ANOVA analysis, correlation analysis, VAR analysis, R/S analysis, and Granger causality test. According to the results, ESG indices are less volatile than conventional ones. Based on the corre- lation analysis, it is concluded that there is a significant direct connection between ESG indices and their traditional counterparts (0.98 for Austria and 0.68 for Ukraine). A substantial level of persistence in Austria’s investment position indicates the possibility of using autoregression models for forecasting. The results of the net investment posi- tion modelling for the case of Austria showed a statistically significant impact of stock market indices on the net investment position. But for the case of Ukraine, this impact is insignificant. This is indirect evidence in favor of poor performance of the Ukrainian stock market. Further development of Ukrainian stock market is required, because Austrian experience showed that stock market can be used as a transmission mechanism in boosting investment position both within conventional approach and ESG. Total global investment flows in 2022 and beyond will be affected by macroeconomic shocks set off by the conflict, pandemic, and increased investor uncertainty, threatened the net investment positions of developed and emerging courtiers. Visvizi and Tokarski (2014) showed that the net international investment position is vulnerable to external shocks.


INTRODUCTION
Investment climate worldwide has changed dramatically as a new wave of COVID-19 and war in Ukraine resulted in a triple crisis of high food and fuel prices combined with the tighter financing conditions. According to the World Investment Report (UNCTAD, 2022), foreign direct investment (FDI) flows recovered to pre-pandemic levels in 2021, hitting nearly USD 1.58 trillion. However, in 2022, a significant decrease is expected. Total global investment flows in 2022 and beyond will be affected by macroeconomic shocks set off by the conflict, pandemic, and increased investor uncertainty, threatened the net investment positions of developed and emerging courtiers. Visvizi and Tokarski (2014) showed that the net international investment position is vulnerable to external shocks.
"Marshal Plan" for Ukraine was presented on the joint Switzerland-Ukraine Recovery Conference (Ukraine Recovery Conference, 2022) in Lugano on the 4-5th of July and prescribed the huge investment support of Ukrainian after war recovery and sustainable economic growth of USD 750 billion (Ukraine Recovery Conference, 2022). Possible improvement of Ukraine's net investment position is expected during three periods of investment resources supply -in 2022, 2023-2025, 2026-2032 with a strong focus on ESG (Environmental, social, and governance) and sustainability criterion.
Stock market architecture can be used as a good transmission mechanism for allocating the investment resources. For Ukraine, it can be a good chance to re-build a stock market infrastructure and reanimate the main functions of narrow, uncompetitive and low-volume market (Plastun, Makarenko, Yelnikova et al., 2018;Plastun, Makarenko & Balatskiy, 2018), create new market indices and benchmarks (both ESG and conventional). Historically, conventional stock market indices describe the situation with investment in the country, and as well as situation in the capital markets (Baumohl, 2012;Thalassinos et al., 2015). Currently, developed countries are trying to improve their investment positions by using ESG investments instead of traditional ones. Comparing both stock market indices is a rather unexplored area (Caporale, 2022 . This pillar is strongly oriented towards the integration of the Ukrainian capital markets into the European financial space, legislation, and benchmarks for long-term ESG investment support of Ukrainian sustainable recovery towards euro integration and estimated in USD 75 billion, 1/10 of the total amount of the investment plan. Austria was chosen for comparison with Ukraine because this country is one of the top 7 EU countries with good dynamics of the net investment position and has strong ESG stock market focus. For instance, dynamics of Austria's net investment position is constantly positive from 2013, and in June 2020 it reached a record high of 15.30% of GDP (Austrian National Bank, 2022). Moreover, the main reason of the Austria's net investment position positive dynamic might be active emission of green and social bonds in the beginning of 2020s on Vienna Stock Exchange. This also may be caused by the rise in the VÖNIX index as one the oldest European ESG index at this exchange (calculated since 2005, Vienna Stock Exchange, 2022). Austrian experience in net investment position improvement and ESG achievements is valuable in the context of Ukraine -EU stock market integration process and strong focus on ESG investment in recovery process of Ukraine.

LITERATURE REVIEW
Despite the importance and crucial role of the net international investment position in the economic development of a country, the academic literature mostly ignores this issue. Results of a bibliometric analysis of the net international investment position in terms of key search queries for the period 2000-2021 (using the built-in tools Scopus from Elsevier, Web of Science from Clarivate Analytics, and Publish or Perish from Google Scholar) showed that most searches are related to the "net international investment position" term. Aspects related to the mutual influence of stock indices on the international investment position are discussed in less than 2.5% of articles (out of the to-tal number). The same applies to the number of citations worldwide. The query "net international investment position AND responsible investment" is characterized by insignificant number of results: only 29 publications. The most significant number of publications on the topic "net international investment position" is presented in Google Scholar. However, even in this database, there is an insignificant number of academic papers related to the influence of traditional and responsible stock market indices on the international invest- Zhong and Liu (2021) used multivariate GARCH models to illustrate dynamic conditional correlations and volatility effects between Chinese and five Southeast Asian stock markets. The authors proposed to include an indicator of China's international investment position in one of the models. However, they further noted that this model was not adequate. Śliwiński (2018) analyzed factors determining changes of the net international investment position of the euro area countries and found that it was largely driven by current account surpluses. Additional factors of influence are exchange rates and prices movements. Knap (2019) evaluated the relationship between the investment position in Poland and population ageing and claimed to found a negative impact of ageing on the international net investment position. Bleys (2008) analyzed the Index of Sustainable Economic Welfare (ISEW) in Belgium, emphasizing the importance of considering its interaction with the net international investment position of the country; Đonlagić and Moskalenko (2020) did the same for the case of The Baltic-Black Sea region countries. There are almost no studies on the influence of stock and responsible indices on Austria's international investment position. An exception is a study by Neher (2015) who studied the impact of RI on German-speaking countries; the author noted that RI positively impacts the country's investment position.
Thus, the impact of the stock market on the net international investment position is not widely discussed in the academic literature, thus it needs further development. This paper aims to fill existing gaps in the net investment position modelling based on information from the stock markets (both conventional and ESG) for the case of Ukraine and Austria. This is important for justifying further Ukrainian stock market recovery directions after the war and more strict regulation on the Austrian stock market.

DATA AND METHODOLOGY
In this study, three arrays of quarterly data both for Ukraine and Austria are used: This paper uses descriptive statistics to analyze the differences between data sets and their behavior, and characteristics; variance analysis to define whether or not differences between data sets are statistically significant; correlation analysis to determine interdependencies between data sets; R/S analysis to analyze data persistence (methodology is similar to Plastun, Makarenko, Yelnikova et al., 2018); Granger causality test to determine whether one data set can be useful in forecasting another; regression analysis and VAR modelling to forecast the net investment position.
The relationship between the net investment position and stock market indices is investigated here by running the following regressions (see equations 1 and 2): where Y t -net investment position at period t; b nnet investment position mean; a 1 -coefficients on conventional stock market index; Conv t -conventional stock market index at period t; ε t -Random error term at period t.
where Y t -net investment position at period t; b nnet investment position mean; a 1 -coefficients on ESG stock market index; Conv t -ESG stock market index at period t; ε t -Random error term at period t.
In this paper, forecasting a country's investment position involves three different models.
The lagged variable of the investment position acts as an independent variable for model 1 (see equation 3). The optimal lag is determined based on the autocorrelation function analysis.
where y t = (y t 1 , y t 2 , ..., y t k ) -is a time series vector; A t is a time-invariant matrix; a 0 is vector of constants; ε t -is a vector of error terms.
where y t = (y t 1 , y t 2 , ..., y t k ) -is a time series vector; A t is a time-invariant matrix; a 0 is vector of constants; y t-i -net investment position at period t-i; Conv t -conventional stock market index at period t; ε t -is a vector of error terms.
where y t = (y t 1 , y t 2 , ..., y t k ) -is a time series vector; A t is a time-invariant matrix; a 0 is vector of constants; y t-i -net investment position at period t-i; Conv t -conventional stock market index at period t; ε t -is a vector of error terms.
These models aim to determine the possibility of using stock market indices (conventional or ESG) as critical factors affecting the investment position.

EMPIRICAL RESULTS
Descriptive statistics for the first differences (Table 1) of ESG and traditional indices indicates lower volatility of ESG indices than their traditional counterparts because the standard deviation of ESG indices, both for Ukraine and Austria, is smaller than traditional indices. Therefore, ESG indices are less risky for investors in both countries. In Austria, ESG indices are also more profitable for investors. However, in Ukraine, traditional indices are more profitable, this can be explained by the fluctuations of the hryvnia (since the ESG index is formed from the prices of shares denominated in euros, and the traditional index is in hryvnia). Based on the ANOVA analysis (Table 2), it can be concluded that the null hypothesis is not rejected (the absence of statistically significant differences between the data sets), so the investment position of the country, the ESG index, and the traditional index behave typically. The next stage is correlation analysis ( Table 3). The results indicate a high degree of correlation between conventional and ESG indices. The value of the correlation index for Austria is 0.98, which shows an almost functional relationship between conventional and ESG indices. For Ukraine, the correlation index is 0.65, which means a close direct connection between the indicators. The correlation between the investment position and conventional and ESG indices for Ukraine is very low (almost absent). The correlation index for the investment position and the traditional index for Austria indicates the presence of inverse, medium-strength relations. To define the possibility of forecasting data based on their previous values, an R/S analysis was conducted (Table 5). R/S analysis of data sets for Ukraine is not possible due to insufficient data. A significant level of persistence of Austria's investment position indicates the "long memory" in data set, which signifies the possibility of using autoregression models for forecasting its values. Fluctuations in the Austrian stock indices are close to random, but demonstrate signs of persistence.
The next stage is the construction of the autocorrelation function (Table 6) for the data series "invest-ment position of the country" (for Ukraine and Austria). As can be seen, lag 1 (the highest values of the autocorrelation function and t-stat parameter) is the most appropriate for the forecasting the investment position based on data from previous values. Table 8 provides the results of regression analysis, the purpose of which is to clarify the fundamental applicability of forecasting a country's investment position based on stock market indices.
Coefficients of determination (0.96 -in the case of Austria and 0.83 -in Ukraine) show that autoregressive models for the first differences are adequate. That is, previous values significantly influence a country's investment position dynamics.

Table 4. Granger causality tests for the case of Ukraine and Austria
Parameter    Taking into account the historical development of the stock market of Ukraine, it can be assumed that as the stock market develops, its role in the country's investment position also increases (especially in the case of traditional indices).

The net investment position (X) and traditional data (Y) ESG data (X) and conventional data (Y)
Next vector autoregressive modelling is used. It includes the following steps: • checking data sets for stationarity (and bringing the time series to a stationary form if necessary); • finding the lags for the model; • performing the Johansen cointegration and the Granger tests.
Stationarity is one of the key assumptions for the vector autoregressive model. To verify the stationarity of the time series, the Dickey-Fuller test is used (for both countries). Note that the calculations are performed using the specialized STATA/IC 12 software. All critical criteria (Table 8) of the Dickey-Fuller test confirm the non-stationarity of the time series data. To solve the issue of the non-stationarity of time series, their transformation by finding their first differences is carried out. After that, checking the analyzed data for stationarity shows their basic acceptability for use in autoregressive models. As can be seen, initial data are non-stationarity. Data sets need to be transformed. For these purposes, the first differences method is used. After transformations, data sets are stationary, VAR modelling can be applied.
One of the key stages of VAR analysis is determining the lags for the models. To do this, the following criteria are used: Final Prediction Error (FPE), Akaike's information criterion (AIC), Hannan -Quinn information criterion (HQIC), and Schwarz Bayesian information criterion (SBIC). Note: * p-value is given in parentheses. Table 9 provides an example of the definition of lags for Austria. The series with the optimal lags (a significant value of p-statistics and the lowest values of information criteria) are marked with the asterisks. The same approach is applied for the Ukrainian data.
The results of choosing the optimal number of lags for both countries are provided in Table 10. The optimal number of lags is selected by assessing the quality of the VAR model using STATA software.   As can be seen, there is no cointegration (cointegration parameters are below critical values).
The next step is VAR analysis. Results are presented in Table 12.
As can be seen, no significant results are found for the case of Model 1 and 2 for both countries. There are no statistically significant interdependencies between the stock market indices and the net in- R/S analysis shows the persistence in Austria's investment position data. ESG indices are more persistent than traditional ones. Regression analysis shows that the previous values can be used to model the net international investment position of a country. These results are confirmed by R/S analysis (high values of the Hurst exponent) and autocorrelation function analysis (the optimal lag is 1).
Austrian stock market traditional indices, as well as ESG ones, have influenced net investment position. Both indices are highly correlated. In this regard, in Austria the stock market can be used as a transmission mechanism to boost investment position both under the conventional approach and ESG. RI investing need to be reinforced into the market infrastructure to improve the net investment position.
Overcoming greenwashing issues and facilitating sustainability and ESG disclosure are a crucial step for 19 companies included in VÖNIX and other publicly listed companies in Austria. Under EU Directive 2014/95/EU and Austrian Commercial Code (UBG), after 2017 they have to disclose ESG matters. But in the light of the latest EU Commission's proposal for a new Corporate Sustainability Reporting Directive and adoption of the EU sustainability reporting standard, these disclosures must be more comprehensive, verified obligatory by independent assurance company, and have digital tags and access.
To clearly articulate ESG and prevent greenwashing, further efforts are needed to implement the EU Taxonomy and EU Taxonomy Climate delegated act (applying from January 1, 2022) into Austrian companies reporting process. Taxonomies unify approaches to ESG disclosure, enhance companies' transparency and reflect the true picture of companies' sustainability performance for key stakeholders.
One more important step for the Austrian ESG segment and ESG stock market indices is the introduction of Regulation on a Voluntary European Green Bond Standard (EU GBS). This scheme is significantly important in the context of new ambitious targets of the Austrian government -full coverage of electricity demand form alternative sources to 2030 and climate neutrality to 2040 as part of the Renewable