Prediction of business cycle of Poland

65 Tkacova, A., Gavurova, B., & Kelemen, K. (2022). Prediction of business cycle of Poland. Journal of International Studies, 15(3), 65-81. doi:10.14254/20718330.2022/15-3/5 Prediction of business cycle of Poland Andrea Tkacova Faculty of Economics, Technical University of Košice, Nemcovej 32, 040 01 Kosice, Slovak Republic andrea.tka[email protected] ORCID 0000-0003-0268-009X Beata Gavurova Faculty of Management and Economics, Tomas Bata University in Zlín, Mostní 5139, 760 00 Zlín, Czech Republic [email protected] ORCID 0000-0002-0606-879X Katarina Kelemen Faculty of Economics, Technical University of Košice, Nemcovej 32, 040 01 Kosice, Slovak Republic [email protected] ORCID 0000-0003-3540-1093 Abstract. The paper is focused on the construction of a new composite indicator intended to predict the economic cycle of Poland and its comparison with the existing CLI used by international institutions such as OECD and Eurostat. In part, this research is also dedicated to monitoring the partial advance cyclical indicators that make up the CLI components and their changes over time. The paper explores 62 qualitative and quantitative economic indicators of Poland and their relationship to the development of monthly GDP at constant prices in three different time periods: 2005 to 2021, 2010 to 2021, and 2016 to 2021. A modified OECD method is used to select the cyclical component of time series using the Hodrick-Prescott filter and subsequently employ cross-correlation of the variables with the cyclical component of GDP. The constructed CLI can predict the evolution of the CLI one month ahead with a cross-correlation level of 0.879 under equal weights and 0.877 under different weights. Research has shown that Received: January, 2022 1st Revision: June, 2022 Accepted: September, 2022 DOI: 10.14254/20718330.2022/15-3/5 Journal of International Studies Scientific Papers © Foundation of International Studies, 2022 © CSR, 2022 Journal of International Studies Vol.15, No.3, 2022 66 there is no significant change in the composition of the CLI for the prediction of the economic cycle of Poland when using the established methodology. Keywords: Business Cycle, Composite Leading Indicator (CLI), GDP, Crosscorrelation, Prediction. JEL Classification: E32, E37. 1. INTRODUCTION Monitoring of the business cycle has been at the forefront of economic research since the 1920s, mainly due to the instability of the economy and the subsequent Great Depression in the 1930s. In the 1950s and the 1960s, the business cycle seemed to be “dead”, but the 1970s and the oil crises brought it back into the spotlight and new economic theories came along trying to explain the causes of its origin (Berge, 2011). A new wave of interest in observation and prediction of the economic cycle occurred at the time in the 21st century that will go down in history as a period full of significant economic shocks that often could not be reliably predicted. The deep negative effects of the financial crisis in 2008, the subsequent wave of the extreme public debts of the European countries, the unexpected global onset of the coronavirus pandemic in 2019, and the war in Ukraine have raised the question of whether and to what extent we are able to predict the possible future development of national economies. Currently, there are various approaches to predicting business cycles using econometric models (Clark & Ravazzolo, 2015; Ferrara, 2015; Barhoumi, 2016; Bjørnland, 2017; Pawęta, 2018; Kliuchnikava, 2022). Some authors use specific general indexes (Gaweł & Głodowska, 2021; Wachira, 2022; Bartos et al. 2021, 2022), however, globally the attention is focused on predictions with the help of composite leading indicators (CLIs). A Composite Leading Indicator (CLI) is an index that aggregates the time series summarising information contained in a number of key short-term economic indicators known to be associated with the business cycle (Zalewski, 2008). The economic cycle is represented by the development of GDP, industrial production index or Composite Coincident Indicator (CCI) (Arnoštová et al., 2011; Bacarić et al., 2016). CLI provides qualitative data on short-term economic movements. CLI draws much attention because it should be able to predict the future states of economic activity – when the economy is going to switch from the expansion phase into the contraction phase or vice versa (Vraná, 2008). Determination of turning points is one of the main benefits of using CLI (Bakarić et al, 2016). It is important to note that CLIs offer information about the expected development of the economy, and thus open a debate on the implementation of decisions in the public or private sector (Monni et al. 2017; Fabuš, 2017). This is a short-term estimate of the future economic situation, but it can point to important economic changes that are the turning points of the economic cycle (Kramoliš, 2015; Mazur, 2017, Döpke et al. 2017). In addition to the prediction itself, CLI can also be used to analyse the performance of economies and thus, to compare their performance within different groups (Saltelli, 2006). The composite leading indicators also possess own disadvantages and limits that are visible in their practical use that is common for the composite indicators (see Djogo & Stanišić, 2016; Roszko-Wójtowicz & Białek, 2016; Šegota et al., 2017). Among the fundamental issues, it is possible to consider the incorrect selection of the reference series or an insufficiently large group of sub-indicators. When using CLI, it is mainly about the availability of the false signals and misinterpretation of the results (Saisana & Tarantola, 2002; Eurostat, 2017). For the economic policy measures derived from the CLI prediction, it is necessary to be cautious and to take into account the impact of several factors, such as the development of CLI partial indicators, the Andrea Tkacova, Beata Gavurova, Katarina Kelemen Prediction of business cycle of Poland 67 overall nature of the economy, the causes of negative developments (problems of the banking sector, public finances, developments in the credit market, the quality of business environment, economic developments in the other countries depending on the openness of the economy, and the other aspects. (Travkina, 2015; Janto-Drozdowska & Majewska, 2016; Fuinhas et al., 2016; Sachpazidu-Wójcicka, 2017; Dobeš et al. 2017; Korsakienė et al. 2017). One of the most important factors that can shift significantly prediction results is the high share of informal economy (Digdowiseiso & Sugiyanto, 2021; Jovovic, 2021; Mishchuk et al., 2018) as well as uncertainty in innovations and knowledge-based activities development (Ebong & Babu, 2020; Oliinyk et al., 2021; Wasiluk & Ginevičius, 2020). It is equally important that some fiscal policy decisions can only be positive for the economy in the short term, but in the long run, they may be counterproductive (Korsakienė et al. 2015; Ciegis et al., 2015; Lajtkepová, 2016; Vochozka et al. 2020, 2021). Due to the advantages of CLI, the national statistical offices, the national banks, as well as the specific enterprises, primarily with an industry orientation, are devoted to explore their construction and subsequently, to analyse these points. CLI possesses an important position in the short-term economic forecasts of the European countries, the United States of America, and Japan. At the international level, the organisations such as the OECD and Eurostat pay a significant attention to them. They prefer a different CLI design approach and hence, they use mainly the same CLI composition for the selected country for a long time. This results in quality prediction capabilities of CLI not being achieved for some economies. For this reason, it is necessary to verify the predictive capabilities of existing CLIs and, if necessary, create new composite pre-heat indicators. For economies, which have undergone the significant changes in the past, it can be assumed that the composition of their CLI will change over time. The transitional economies can be considered such countries. For the purposes of this paper, Poland was selected as the country under study due to the insufficient predictive ability of CLI by the OECD and Eurostat and this is analysed in more detail in the given paper. The aim of the present contribution is to create an own composite leading indicator for the business cycle of Poland and compare it with the existing CLIs. The research question is also raised whether there really is a change in the composition of the CLI in the case of Poland over time. 2. LITERATURE REVIEW The economists at the national and international level are engaged in the construction of the CLI for the economic cycle of Poland. The best-known international methodologies for the construction of Poland's CLI are represented by the OECD and Eurostat procedures. Their modifications can be seen in the studies of the authors such as Bandholz (2005), Zalewski (2009), Jakubíková et al. (2014), and Vraná (2018). The methodology of the OECD and Eurostat is based on the growth cycle, while the time series can be divided into the random, trend, seasonal, and cyclical components (Trimbut, 2006). For further investigation, the cyclical component is selected from the time series. The first important step is the selection of a reference series that represents the economic cycle of the given economy. The studies offer several options for a selection of the reference series for Poland. The monthly index of the industrial production (or manufacturing production) is the most commonly used measure of economic activity (Bandholz, 2005). The first reason is that it is available promptly and on a monthly basis in contrast to GDP. Secondly, it constitutes the most cyclical subset of the whole economy. Moreover, for many countries it was found that cyclical profiles of GDP and IIP are strongly related. An obvious disadvantage of employing GDP instead of IIP is that GDP is very often revised by the Central Statistical Office and it is a subject to significant changes (OECD, 2006). According to the OECD methodology, the IIP was applied to construct CLI until 2012 for Poland and from 2012 it was a monthly time series of GDP at constant prices (Fulop & Gyomai, 2012). GDP is also preferred by Eurostat (2017), Bandholz (2005), Zalewski (2008), and Jakubíková et al. (2014). On the other hand, authors such as Artis et al. (2004) and Vraná (2018) prefer the use of IIP. In Journal of International Studies Vol.15, No.3, 2022 68 their opinion, GDP for Poland shows too little cyclical variation and is thus not the appropriate measure for monitoring business cycle fluctuations. A brief comparison of the elementary differences in the construction of the OECD CLI and Eurostat composite indicators is presented in Table 1. Table 1 Comparison of international methodologies for CLI construction Criteria Eurostat OECD Type of business cycle Grow cycle Grow cycle Countries Chosen countries of European Union, Eurozone Members of OECD, a few nonmember countries Trend determination Hodrick Prescott Filter Christiano Fitzgerald Filter Reference series GDP Index of industrial production (to 2012) GDP (since 2012) Relationship between reference series and indicators Cross correlation Cross correlation Type of data Qualitative data Quantitative and qualitative data Period of data Monthly data Monthly data Wages of components Diggerent wages for all components Same wages for all components Source: Authors’ results. The prediction possibilities for the cyclical development of Poland's economy are discussed in the studies by Drozdowicz-Bieć (2001), Matkowski (2002), Nilsson (2006), Garczarczyk & Skikiewicz (2011), Lenart et al. (2016) or Mazur (2017). Another author Bandholz (2005) applies GDP as the reference series and thus, he creates the CLI for Poland and Hungary from quarterly data for the period 1994 to 2004. Employing the linear and non-linear dynamic factor modelling approaches, he finds for the both countries that a parsimonious specification that combines the national business cycle indicators, the series reflecting trade volumes, and the supranational business expectations making for the most reliable business cycle leaders. The composite leading indicators significantly cause a GDP growth, while the estimated Markovswitching probabilities of being in a recessionary state agree well with a priori determined cycle chronologies. The resulting CLI for Poland is composed of a system of the equal and different weights and it contains the six components. Zalewski (2009) applies the modified form of the OECD methodology and thus, he chooses monthly IIP data for the period 1992 to 2007 as the reference series. Based on the analysis of the 15 economic indicators of Poland, he compiles the CLI, which consists of six resulting components. He compares the constructed CLI with the development of ESI and Matkowski (2002), evaluating that his constructed CLI indicates fewer false signals and therefore, it provides a better short-term prediction. Jakubíkova et al. (2014) explored the development of the economic cycle of the V4 countries with a reference series of GDP at constant prices. Using the analysis of the quarterly data for the period 2005 to 2011, they pointed out the shortcomings in the OECD predictive ability for Poland. In the explored period, the highest value of the OECD CLI cross-correlation was 0.489, while Eurostat ESI stood at a level of 0.798. Subsequently, according to the modified OECD methodology, a CLI was compiled for Poland too, and it consisted of the six components also. For the V4 countries, the CLI has been shown to have a different composition for each country. Vraná (2018) applied an innovative approach to the construction of the CLI, which consists in including variables from other countries among the selected cyclical indicators. To compile CLI of Poland, it follows the development of indicators in Austria, Germany, the Czech Republic and Slovakia for the period 1996 to 2016 with the reference series of GDP. He argues that the economies of some countries are Andrea Tkacova, Beata Gavurova, Katarina Kelemen Prediction of business cycle of Poland 69 small and their economic situation is often related to the development of the business cycle in other countries. According to her study, the resulting CLI of Poland actually achieves the best predictive capabilities when incorporating indicators from the monitored countries. The available CLIs for Poland were compared based on selected criteria such as a CLI composition, lead size and CLI cross-correlation value (Table 2). Table 2 Comparison of the construction of the composite indicators for Poland Author Observed period Indicator construction Time period advance Crosscorrelation value Bandholz (2005) 1994-2004 (quarterly data) CLI (reference series GDP): Describes demand tendency in manufacturing, Nominal effective exchange rate, Business expectations for Western Europe 1 Q 0.75 Zalewski (2009) 1992-2007 (monthly data) CLI (reference series IPP): Narrow Money (M1) Index 2000=100 SA, Manufacturing industry (Selling prices, Future tendency), Manufacturing industry (Production Tendency), Short-term interest rates, Manufacturing industry (Production, Future Tendency), Net trade in goods (value) in billions of US dollars SA - 0.788 Jakubíková et al. (2014) 2005-2011 (quarterly data) CLI (reference series GDP): Production in manufacturing industry, 2005 = 100 Monetary aggregate M1, 2005 = 100 Warszawski Indeks Giełdowy, 2005 = 100, Industry turnover, (intermediate product and capital goods), domestic market, 2005 = 100 Indicator of confidence in the construction industry 2 Q 0.886 Vraná (2018) 1996-2016 (monthly data) CLI (reference series GDP): 9 international components (Austria, Czech Republic, Germany, Slovakia) and 6 national components: Business tendency surveys (construction), Share Prices, Business tendency surveys (manufacturing), International Trade (Imports, goods), Business tendency surveys (servicesbusiness situation), Business tendency surveys (services - confidence indicators) 3 M 0.790 OECD* (2022) 2005-2021 2010-2021 2016-2021 (monthly data) CLI (reference series GDP): Real effective exchange rates - CPI Based (2015=100) inverted 3-month WIBOR (% per annum) inverted Manufacturing - Production: tendency (% balance) Job vacancies: unfilled (number) Production of coal (tonnes) 10 M 9 M 9 M 0.442 0.578 0.588 Eurostat* (2022) 2005-2021 2010-2021 2016-2021 (monthly data) CLI (reference series GDP): Confidence indicator in industry, Confidence indicator in the service, Consumer confidence indicator, Confidence indicator in construction, Confidence indicator in retail 1 M 0 M 0 M 0.756 0.811 0.921 Note: * own calculations Source: Authors’ results Journal of International Studies Vol.15, No.3, 2022 70 In the case of CLI OECD and ESI Eurostat, the values were calculated by the authors of the paper. By comparing the observed studies, the difference in the predictive abilities of CLI is noticeable. The reason is mainly differences in the CLI calculation methodology, but also in the type and periodicity of monitored variables and the length of the time series. The predictive capabilities of the OECD and Eurostat CLIs, which were calculated by the authors of this study from three different length time series with unchanged CLI composition, are interesting. In the case of the OECD, a prediction period of up to 9 or 10 months was found, but with a low cross-correlation value of 0.442 to 0.588. This means that with this composition, the CLI gives many false signals about the changes in the Polish business cycle. On the contrary, the ESI showed a high correlation value from 0.756 to 0.921, but at the time of the coincidence or a month in advance. 3. METHODOLOGY The main goal of the paper is the construction of a new composite leading indicator (CLI) designed to predict the development of the business cycle in Poland. A partial objective is to determine whether the length of the time series significantly affects the groups of the cyclical leading indicators that can form CLIs. For creation of CLI, a modified OECD methodology is employed. It is based on a construction with the growth cycle (De Vroey & Pensieroso, 2006). This is more appropriate to apply in the case of the transitional economies such as Poland (Czesaný et al., 2007). The monthly GDP indicator at constant prices is selected as the reference series that is generally considered the broadest indicator of economic activity (Czesány & Jeřábková, 2009a). Only its cyclical component is used in the analyses (Czesány & Jeřábková, 2009b; Astolfi et al., 2016). The indicators, whose relation to GDP is monitored include, the 62 quantitative and qualitative indicators from the fields of industry, services, retail, construction, foreign trade, labor market, monetary aggregates, stock indices, confidence indicators, consumer expectations as well as the GDP components themselves. The data sources are the databases of the OECD, Eurostat and the Polish Statistical Office. All the available time series from January 2005 to November 2021 with monthly periodicity are included in the analysis. The relationship of the variables to GDP in the periods 2005 to 2021, 2010 to 2021, and 2016 to 2021 is monitored. The reason is the results of the published research for Slovakia, which confirmed there is a change in the composition of the CLI over time or that the length of the observed period affects the composition of the CLI. The recommendation is to monitor changes in the composition of the CLI at least at the five-year intervals, or after major economic changes in the country and thereby, to minimise the number of the false signals (Tkáčová & Kišová, 2018). The time series is initially seasonally adjusted by the method of balancing employing the seasonal indices and then, the trend is removed applying the Hodrick-Prescott filter (HP filter). The HP filter is a commonly employed tool for detrending. It is a most favourable extractor of a trend that is stochastic but moves smoothly over time and is uncorrelated with the cycle (Kovacic & Vilotic, 2017). For t=1,2,3… the trend component Y* is computed, and  is chosen to minimise: ∑(𝑌 𝑡 𝑇 𝑡=1 − 𝑌 𝑡∗)2+ λ ∑[(𝑌 𝑡+1 ∗− 𝑌 𝑡∗ 𝑇−1 𝑡=2 ) − (𝑌 𝑡∗− 𝑌 𝑡−1 ∗)]2 (1) To get optimal results for detrending, it has been suggested to choose λ=1600 for quarterly data and λ =14 400 for monthly data (Schilcht, 2005). An advantage of the HP method is that no restriction on the length of time series is imposed. Nevertheless, there is a requirement that before proceed with the HP filter one should seasonally adjusted each series. The trend itself is not very interesting in the analysis of cyclical Andrea Tkacova, Beata Gavurova, Katarina Kelemen Prediction of business cycle of Poland 71 behaviour. Therefore, the rest of study was done with cyclical components of each series (Nilsson & Brunet, 2006). The cross-correlation in a way of the Pearson's correlation coefficient with the two forward shifts and the five backward shifts is applied in order to determine the relationship between the variables. According to the value of the correlation coefficient and the size of the advance, it is possible to create groups of cyclical and anticyclical indicators according to the pattern of Table 3. Table 3 Criteria for identification of the cyclical behaviour of the indicators for Poland Indicator type The highest absolute crosscorrelation value Time to reach the highest absolute cross-correlation value Cyclic Early > 0,55 (t+1, t+2) Overdue > 0,55 (t-5, t-1) Concurrent > 0,55 t Anticyclical ≤ 0,55 not significant Source: Kľúčik, Haluška (2008). A group of the leading indicators is a subject to a more detailed analysis for the compilation of Poland's CLI. Subsequently, the selection and scoring method is applied, which economic and statistical significance and statistical quality are evaluated according to. The maximum number of the points that the variables can reach is given in Table 4. Table 4 Scoring criteria for the selection of the leading cyclical indicators Economic significance (maximum 10 points) Statistical significance (maximum 30 points) Statistical quality (maximum 10 points) Economic interpretation in relation to the business cycle 10 points Pearson's correlation coefficient 15 points Time availability 5 points Number of advance months 15 points Update 5 points Source: Authors’ results Due to the different units of the partial indicators, their normalised values obtained by using the standardization method (OECD, 2008) are used in the composition of the CLI. A system of the equal and different weights is employed. The relationship for calculating equal weights looks like as follows: 𝜔𝑖=1 𝑛 (2) where i stands for the weight of ith component a n is the number of the partial indicators entering the CLI. The absolute values of the correlation coefficients are applied in order to determine the different weights. The relationship for calculation of the different weights is expressed subsequently: 𝜔𝑖= 𝑟𝑖 ∑𝑟𝑖 𝑛 𝑖=1 (3) where i stands for the weight of ith component, r is the absolute value of the correlation coefficient of ith component at the time of advance and n is the number of the partial indicators entering the CLI. The sum of the normalised values multiplied by the weights creates the basement for construction of the CLI assembly equation. 𝐶𝐿𝐼𝑡= ∑𝜔𝑖 𝑛 𝑖=1 ∗ 𝑦𝑖,𝑡 (4) 𝜔 – a variable weight value, y – a value of the normalised cyclic component of the variable at t time, Journal of International Studies Vol.15, No.3, 2022 72 n – a number of the partial indicators entering the CLI. 4. EMPIRICAL RESULTS AND DISCUSSION The basic step for the creation of CLI for Poland is characterised by the selection of a reference series that represents the economic cycle. Through the analysis of the relationship between the cyclical component of GDP and PPI, it was demonstrated that PPI shows a month advance with a cross-correlation value of 0.789 for the period 2005 to 2021 and 0.827 for the period 2016 to 2021. This means that the PPI represents the leading indicator of GDP and thus, it is able to form one of the components of the CLI. It is not appropriate to select it as a reference line. In the case of Poland, it is more suitable to use the cyclical component of GDP. Selection of the cyclical indicators for Poland According to the described methodology, the cyclical relationship of all the 62 selected variables to GDP was analysed. Through calculation of the cross-correlations of the cyclical components associated with the variables of the cyclical component of GDP, the 17 indicators were selected that demonstrated the properties of the leading cyclical indicators according to the definition given in the methodological part of the paper. The highest cross-correlation value was achieved to the left side of t and at the same time, the second highest cross-correlation value was greater than 0.55. Table 5 demonstrates the results. Table 5 Advance indicators Indicator 2005–2021 2010–2021 2016–2021 advance correlation advance correlation advance correlation Industrial production index 1 M 0.798 1 M 0.802 1 M 0.827 Stock price index 2 M 0.524 2 M 0.533 1 M 0.664 Total industry production index 1 M 0.807 1 M 0.813 1 M 0.838 Total production in the processing industry index 1 M 0.775 1 M 0.774 1 M 0.796 Total production in the processing industry index – intermediate product 1 M 0.732 1 M 0.756 1 M 0.761 Total investment goods production index 1 M 0.745 1 M 0.751 1 M 0.757 Total retail volume index 1 M 0.595 1 M 0.622 1 M 0.745 Production in the processing industry 1 M 0.667 1 M 0.723 1 M 0.813 Consumer industry employment 1 M 0.599 1 M 0.665 1 M 0.747 Processing industry confidence indicator 1 M 0.714 0 M 0.786* 0 M 0.857* Construction industry employment 1 M 0.619 1 M 0.719 1 M 0.796 Retail business situation 1 M 0.706 1 M 0.750 1 M 0.801 Retail employment 1 M 0.702 1 M 0.805 1 M 0.834 Order intention or demand in retail 1 M 0.708 1 M 0.760 1 M 0.811 Services – demand development 1 M 0.760 1 M 0.788 1 M 0.862 Services – employment 1 M 0.686 1 M 0.766 1 M 0.813 Dow Jones euro stoxx 50 Price Index 1 M 0.483* 1 M 0.479* 1 M 0.667 * the indicator is assigned to the advance indicators according to the advance period length and not the crosscorrelation value Source: Authors’ results. Andrea Tkacova, Beata Gavurova, Katarina Kelemen Prediction of business cycle of Poland 73 Three time periods were monitored in order to verify the assumption that there is a change in the leading cyclical indicators throughout the time that represent potential components of the CLI. In the case of Poland, only a small change in the nature of these indicators was confirmed. Out of the 62 variables, 17 indicators were included among the leading cyclical indicators. Subsequently, only the 2 out of the 17 indicators changed their nature as a leading cyclical indicator. The first one was an indicator of confidence in the manufacturing industry, which over time changed from a leading indicator to a concurrent indicator, and the second was the Dow Jones Euro Stoxx 50 price index that acquired the characteristics of a leading indicator over time. For the 15 indicators, it has been proven that they show a better quality lead with a shorter time period, and thus the value of their correlation coefficient increases. Overall, it can be assumed that in the case of Poland, there is no significant change over a time period in the group of the leading cyclical indicators. According to the established methodology, it is possible to compile a high-quality CLI therefore, whose composition would meet the prediction of the economic cycle of Poland even for several years. However, with shorter time periods, the value of the cross-correlation grew, which means that the selected indicators are suitable for creating a CLI for the current prediction. For this reason, the subsequent investigation is devoted to the time series for the 2016 to 2021 period. For a detailed selection of the variables that create the CLI, a selection and a scoring method is applied for all the 17 lead indicators (Table 6), while the maximum value, which the indicator can reach, is 50 points (Kľúčik, 2009b). The variables from the same economic area, which achieved the highest number of points after an application of a selection and a scoring method, were eligible for further analysis. This reduced the selection down to the 10 leading cyclical indicators. Table 6 Results of the selection and scoring method for the advance cyclical indicators Indicator Maximum points Indicator Maximum points Total industry production index 45 Total retail volume index 40 Industrial production index 44 Retail employment 39 Total investment goods production index 43 Order intention or demand in retail 39 Stock price index 42 Processing industry confidence indicator employment 39 Total production in the processing industry index – intermediate product 42 Processing industry confidence indicator 38 Total production in the processing industry index 41 Services – demand development 38 Total production in the processing industry 41 Dow Jones euro stoxx 50 Price Index 38 Retail business situation 41 Processing industry confidence indicator 25 Services – employment 41 Source: Authors’ results Construction of a composite lead indicator for Poland During the construction of the CLI, there were the 7 options for the composition of the CLI with each additional possibility having one less indicator with the lowest number of the points. CLI 10 (at the same weights) or CLI A (at different weights) were composed of the 10 indicators and CLI 4 of the 4 indicators that represented the minimum number. CLI numerical designation is applied for the same indicator weights and a letter designation is used employed for the different CLI component weights. Table 7 demonstrates the results of the cross-correlations with the advance volume. Journal of International Studies Vol.15, No.3, 2022 80 Šegota, A., Tomljanović, M., & Huđek, I. (2017). Contemporary approaches to measuring competitiveness – the case of EU member states. Journal of Economics and Business, 35 (1), 123-150. https://doi.org/10.18045/zbefri.2017.1.123. Tkáčová, A., & Kišová, V. (2017). Zmeny zloženia kompozitného predstihového indikátora Slovenska v čase. Politická Ekonomie, 65(5), 583-600. https://doi.org/10.18267/j.polek.1163. Travkina, I. (2015). Export and GDP Growth in Lithuania: Short-run or Middle-run Causality? Entrepreneurship and Sustainability, Issues 3(1), 74-84. https://doi.org/10.9770/ jesi.2015.2.4(7). Trimbur, T. M. (2006). Detrending economic time series: a Bayesian generalization of the Hodrick-Prescott filter. Journal of Forecasting, 25(4), 247-273. doi:10.14254/2071789x.2015/8-2/10. Vochozka, M., Kalinová, E., Gao, P. & Smolíková, L. (2021). Development of copper price from July 1959 and predicted development till the end of year 2022. Acta Montanistica Slovaca, 26(2), 262-280. https://doi.org/10.46544/AMS.v26i2.07 Vochozka, M., Horák, J., Krulický, T. & Pardal, P. (2020). Predicting future Brent oil price on global markets. Acta Montanistica Slovaca, 25(3), 375-392. https://doi.org/10.46544/AMS.v25i3.10 Vraná, L. (2018). On extending composite leading indicators by international economic series. Statistika. 98(2). 113-13 Wachira, E.W. (2022). Analysis of Austria's Entrepreneurial Ecosystem Based on the Gei Approach. International Journal of Entrepreneurial Knowledge, 10(1), 123-136. https://doi.org/10.37335/ijek.v10i1.156 Wasiluk, A., & Ginevičius, R. (2020). Pro-innovative motives for establishing cooperation by enterprises: An empirical study in Poland. Economics and Sociology, 13(2), 258-278. https://doi.org/10.14254/2071-789X.2020/13-2/17 Zalewski, K. (2009). Forecasting Turning Points with Composite Leading Indicators - the Case of Poland. Ekonomia Journal, 24, 61-9. Andrea Tkacova, Beata Gavurova, Katarina Kelemen Prediction of business cycle of Poland 81 ANNEX Figure 1. Development of the cyclical component of GDP and CLI with the equal and different weights Source: Authors’ results Figure 2. Comparison of the development of the cyclical component of GDP, CLI 5 (the equal weights), CLI OECD, ESI Source: Authors’ results -8,00 -6,00 -4,00 -2,00 0,00 2,00 4,00 Jan-2016 Mar-2016 May-2016 Jul-2016 Sep-2016 Nov-2016 Jan-2017 Mar-2017 May-2017 Jul-2017 Sep-2017 Nov-2017 Jan-2018 Mar-2018 May-2018 Jul-2018 Sep-2018 Nov-2018 Jan-2019 Mar-2019 May-2019 Jul-2019 Sep-2019 Nov-2019 Jan-2020 Mar-2020 May-2020 Jul-2020 Sep-2020 Nov-2020 Jan-2021 Mar-2021 May-2021 Jul-2021 Sep-2021 Nov-2021 Cyclical component of GDP CLI 5 CLI F -5,00 -4,00 -3,00 -2,00 -1,00 0,00 1,00 2,00 -8,00 -6,00 -4,00 -2,00 0,00 2,00 4,00 Jan-2016 Mar-2016 May-2016 Jul-2016 Sep-2016 Nov-2016 Jan-2017 Mar-2017 May-2017 Jul-2017 Sep-2017 Nov-2017 Jan-2018 Mar-2018 May-2018 Jul-2018 Sep-2018 Nov-2018 Jan-2019 Mar-2019 May-2019 Jul-2019 Sep-2019 Nov-2019 Jan-2020 Mar-2020 May-2020 Jul-2020 Sep-2020 Nov-2020 Jan-2021 Mar-2021 May-2021 Jul-2021 Sep-2021 Nov-2021 cyclical component of GDP CLI OECD ESI CLI 5