269 ECONOMIC AND BUSINESS REVIEW | VOL. 20 | No. 2 | 2018 | 269-302 SLOPOL10: A MACROECONOMETRIC MODEL FOR SLOVENIA KLAUS WEYERSTRASS 1 Received: September 26, 2017 REINHARD NECK 2 Accepted: January 22, 2018 DMITRI BLUESCHKE 3 BORIS MAJCEN 4 ANDREJ SRAKAR 5 MIROSLAV VERBIČ 6 ABSTRACT: This paper describes the SLOPOL10 model, a quarterly macroeconometric model of the Slovenian economy to be used for forecasting macroeconomic development and simulating alternative policy measures. The model is of the Cowles Commission type and is estimated using the cointegration approach, thus combining the long-run equilibrium and the short-run adjustment mechanism. It contains behavioural equations and identities for the goods market, the labour market, the foreign exchange market, the money market, and the government sector. Estimation of behavioural equations for Slovenian aggregates is based on data starting in 1995. The model combines Keynesian and neoclassical elements. The Keynesian elements determine the short and medium-run solutions in the sense that the model is demand-driven and persistent disequilibria in the goods and labour markets are possible. The supply side incorporates neoclassical features. Static and dynamic ex-post simulations show that the model can reasonably reproduce past development and is therefore suited for prediction and policy evaluation, especially for fiscal policy design and optimal control experiments. Keywords: SLOPOL10 model, macroeconometric models, fiscal policy design, optimal control experiments, Slovenia JEL Classification: E01, B23 DOI: 10.15458/85451.62 1 Alpen-Adria-Universität Klagenfurt, Department of Economics, Klagenfurt, Austria & Institute for Advanced Studies, Macroeconomics and Public Finance Group, Vienna, Austria, e-mail: klaus.weyerstrass@aau.at 2 Alpen-Adria-Universität Klagenfurt, Department of Economics, Klagenfurt, Austria, e-mail: reinhard. neck@uni-klu.ac.at 3 Alpen-Adria-Universität Klagenfurt, Department of Economics, Klagenfurt, Austria, e-mail: dmitri. blueschke@aau.at 4 Institute for Economic Research, Ljubljana, Slovenia, e-mail: majcenb@ier.si. 5 Institute for Economic Research, Ljubljana, Slovenia & University of Ljubljana, Faculty of Economics, Ljubljana, Slovenia, e-mail: srakara@ier.si 6 Corresponding author, University of Ljubljana, Faculty of Economics, Ljubljana, Slovenia & Institute for Economic Research, Ljubljana, Slovenia, e-mail: miroslav.verbic@ef.uni-lj.si 270 ECONOMIC AND BUSINESS REVIEW | VOL. 20 | No. 2 | 2018 1. INTRODUCTION This paper presents SLOPOL10, a medium-sized macroeconometric model for the small open economy of Slovenia. We document the theory behind the model blocks, the equations, and formal tests of the ability of the model to replicate the trajectories of the endogenous variables in an ex-post simulation. The Slovenian economy, although small, is of interest for the following reasons: First, it was part of the Yugoslav economy, a centrally planned economy with a unique system of workers’ self-management, until the dissolution of Yugoslavia. Second, Slovenia has developed towards a parliamentary democracy and a capitalist economy much faster than any other of the successor states of Yugoslavia. In particular, it became a member of the European Union in 2004 and, as the first former communist country, joined the Euro Area in 2007, which at the time was regarded as a major achievement. Third, the Slovenian economy is one of the small open economies within the Euro Area; hence its economic policy problems may also be of interest to other economies of that type. For example, difficulties resulting from the particular policy architecture of supranational monetary policy versus a national fiscal policy occur not only in Slovenia but also in several other members of the Euro Area. Finally, Slovenia was hit very hard by the Great Recession and the ensuing sovereign debt crisis but managed to return to satisfactory growth relatively fast recently, so it can be regarded as a model for dealing with business cycles. If we want to explain economic developments in a country like Slovenia, and even more so if we want to design economic policies for such a country, a model of the Slovenian economy is required. Such a model shall serve as a tool for forecasting macroeconomic developments over the short and medium run and for evaluating alternative policies aimed at influencing the business cycle, stabilizing unemployment and inflation, and enhancing growth and employment in Slovenia. Several modelling strategies are available for building a macroeconomic model which can fulfil these requirements. If a model builder believes in neoclassical or New Keynesian macroeconomic theory, a Dynamic Stochastic General Equilibrium (DSGE) model will be his/her choice. If, on the other hand, theories are distrusted and a “data-only” approach is preferred, a vector autoregression (VAR) model will be chosen. Here we follow a more traditional modelling approach and opt for an econometric model of the Cowles Commission type. These models compromise between the theory-first and the empirics-first approaches; they must be based on sound theoretical foundations and estimated using real data of the economy under consideration. Several models of this type have been estimated before by members of the present team of authors (Verbič 2005, 2006, Weyerstrass et al. 2007); here we follow this tradition. To build such a model, it is important to have available a data base with sufficiently long time series to provide reliable estimates. For former communist countries like Slovenia, this poses a problem: data before 1991, when the country gained independence, are based on communist accounting rules and are not comparable to those of later years. Even for the early years of the transition process many data (especially those from national income accounting) are of dubious quality. Therefore estimation of behavioural 271 K. WEYERSTRASS, R. NECK, D. BLUESCHKE, B. MAJCEN, A. SRAKAR, M. VERBIČ | SLOPOL10 ... equations for Slovenian aggregates has to be based on data starting in 1995 or later. In order to obtain estimations with sufficient degrees of freedom, an econometric model for Slovenia has to use quarterly or – where available – monthly or even higher-frequency data. Here we describe a quarterly macroeconometric model called SLOPOL10, which is a revised and updated version of a series of models which we have built since the late 1990s, with increasing degrees of sophistication and reliability. These models have been used for various purposes of forecasting and especially evaluating alternative policies, where simulation and optimization experiments were conducted to arrive at politically relevant insights and policy recommendations (see, e.g., Neck et al. 2011). Of particular importance with respect to Slovenia’ s position in the European Union are evaluations of its fiscal policies as the country has to fulfil the requirements of the EU Stability and Growth Pact (see Blueschke et al. 2016). Like every structural econometric model, the SLOPOL10 model may be subject to the famous Lucas critique. Lucas (1976) argued that the relations between macroeconomic aggregates in an econometric model should differ according to the macroeconomic policy regime in place. In this case, the effects of a new policy regime cannot be predicted using an empirical model based on data from previous periods when that policy regime was not in place. Sargent (1981) argues that the Lucas critique is partly based on the notion that the parameters of an observed decision rule should not be viewed as structural. Instead, structural parameters in Sargent’s conception are just “deep parameters”, such as preferences and technologies. These parameters would be invariant, even under changing policy regimes. Providing for such “deep parameters” requires a different class of macroeconomic models, namely Computable General Equilibrium (CGE) or DSGE models. We take the Lucas critique into account to a certain extent by following the so- called London School of Economics tradition initiated by Sargan (1964). According to this approach, economic theory guides the determination of the underlying long-run specification while the dynamic adjustment process is derived from an analysis of the time series properties of the data series. Error correction models involving cointegrated variables combine the long-run equilibrium and the short-run adjustment mechanism. 2. MODEL DESCRIPTION SLOPOL10 (SLOvenian economic POLicy model, version no. 10) is a medium-sized macroeconometric model of the Slovenian economy. In its current version, SLOPOL10 consists of 75 equations, 23 of which are behavioural equations and 52 identities. In addition to the 75 endogenous variables, the model contains 41 exogenous variables. A list of the variables used in the SLOPOL10 model can be found in Table A1 in the Appendix. The model is constructed in order to allow for forecasts and policy simulations over the near future. Statistical tests will be presented that show the performance of the model in the past. In our view, these tests show that the model exhibits acceptable quality for such uses. Improvements in the light of new data will be continually made when using the model for these purposes. 272 ECONOMIC AND BUSINESS REVIEW | VOL. 20 | No. 2 | 2018 The behavioural equations were estimated with the software program EViews, using quarterly data for the period 1995q1 to 2015q4. Data for Slovenia and for Euro Area aggregates as well as the oil price were taken from the Eurostat database, and those for world trade came from the CPB Netherlands Bureau for Economic Policy Analyses. The model contains behavioural equations and identities for the goods market, the labour market, the foreign exchange market, the money market, and the government sector. Rigidities of wages and prices are taken into account. The model combines Keynesian and neoclassical elements, the former determining the short and medium-run solutions in the sense that the model is demand-driven and persistent disequilibria in the goods and labour markets are possible. In the following, the model equations are described verbally. A diagram of the building blocks of the model is given in Figure 1. Figure 1: SLOPOL10 – Building Blocks The supply side incorporates neoclassical features. In accordance with the approach applied by the European Commission for all EU Member States (Havik et al. 2014), potential output is determined by a Cobb-Douglas production function with constant returns to scale. It depends on trend employment, capital stock and autonomous technical 4 The supply side incorporates neoclassical features. In accordance with the approach applied by the European Commission for all EU Member States (Havik et al. 2014), potential output is determined by a Cobb-Douglas production function with constant returns to scale. It depends on trend employment, capital stock and autonomous technical progress. Trend employment is defined as the labour force minus natural unemployment, the latter being defined via the non-accelerating inflation rate of unemployment (NAIRU). In line with the literature on production functions as well as international practice in macroeconometric modelling, the elasticities of labour and capital were set at 0.65 and 0.35 respectively. These elasticities correspond approximately to the shares of wages and profits respectively in national income. The NAIRU, which approximates structural unemployment, is estimated by applying the Hodrick-Prescott (HP) filter to the actual unemployment rate. For forecasts and simulations, the structural unemployment rate is then extrapolated with an autoregressive (AR) process. Capital stock enters the determination of potential GDP not with its trend level but with its actual one. Several steps are required to determine technical progress. First, ex-post total factor productivity (TFP) is calculated as the Solow residual, i.e. that part of the change in GDP that is not attributable to change in the production factors of labour and capital, weighted with their corresponding production elasticities. In a second step, the trend of technical progress is then determined by applying the HP filter, in a procedure similar to the NAIRU. For simulations and forecasts, the trend of the TFP is explained in a behavioural 273 K. WEYERSTRASS, R. NECK, D. BLUESCHKE, B. MAJCEN, A. SRAKAR, M. VERBIČ | SLOPOL10 ... progress. Trend employment is defined as the labour force minus natural unemployment, the latter being defined via the non-accelerating inflation rate of unemployment (NAIRU). In line with the literature on production functions as well as international practice in macroeconometric modelling, the elasticities of labour and capital were set at 0.65 and 0.35 respectively. These elasticities correspond approximately to the shares of wages and profits respectively in national income. The NAIRU, which approximates structural unemployment, is estimated by applying the Hodrick-Prescott (HP) filter to the actual unemployment rate. For forecasts and simulations, the structural unemployment rate is then extrapolated with an autoregressive (AR) process. Capital stock enters the determination of potential GDP not with its trend level but with its actual one. Several steps are required to determine technical progress. First, ex-post total factor productivity (TFP) is calculated as the Solow residual, i.e. that part of the change in GDP that is not attributable to change in the production factors of labour and capital, weighted with their corresponding production elasticities. In a second step, the trend of technical progress is then determined by applying the HP filter, in a procedure similar to the NAIRU. For simulations and forecasts, the trend of the TFP is explained in a behavioural equation. In accordance with the endogenous growth literature, technical progress is influenced by the share of people with tertiary education in the labour force. In addition, trend TFP is influenced by the real investment ratio, i.e. gross fixed capital formation over GDP . As a third factor, lagged real government spending on research and development (R&D) is included in the TFP equation. On the demand side, the consumption of private households is explained by a combination of a Keynesian consumption function and a function in accordance with the permanent income hypothesis and the life cycle hypothesis. Thus, private consumption depends on current disposable income and on the long-term real interest rate, the latter entering the consumption equation with a negative sign. Real gross fixed capital formation is influenced by the change in real disposable income (more or less in accordance with the accelerator hypothesis) and by the user cost of capital, where the latter is defined as the real interest rate plus the depreciation rate of capital stock. Changes in inventories are treated as exogenous in the SLOPOL model, as in many macroeconomic models in use around the world. Real exports of goods and services are a function of the real exchange rate and foreign demand for Slovenian goods and services. Foreign demand is approximated by the volume of world trade. The real exchange rate is meant to capture the competitiveness of Slovenian companies on the world market. Real imports of goods and services depend on domestic final demand and on the real exchange rate. A real appreciation of the Slovenian currency (the Slovenian tolar until the end of 2006 and the euro following Slovenia’s entry into the Euro Area on 1 January 2007) makes Slovenian goods and services more expensive on the world markets. On the other hand, foreign products become relatively cheaper; hence domestic production is substituted by imports. Thus a real appreciation stimulates imports while having a negative effect on exports. Even when Slovenia is part of the Euro Area, its real exchange rate can, of course, still appreciate or depreciate, not only against 274 ECONOMIC AND BUSINESS REVIEW | VOL. 20 | No. 2 | 2018 other currencies but also against other Euro Area countries due to inflation differentials. On the labour market, both labour demand and supply are divided into the main age group (15 to 64 years) and older people (65 years and above). The labour demand of companies (actual employment) is modelled via the employment rates of the two age groups, i.e. employment as a share of the relevant age group in the total population. Both equations were estimated as T obit models, the employment rates being limited to lying between 0 and 0.9 (15 to 64 years) and between 0 and 0.5 (65 years and above). Both employment rates are influenced positively by real GDP and negatively by the real net wage and additionally by the wedge between the gross and the net wage. The idea behind the latter is that increases in the tax wedge are borne partly by employers and partly by employees. Rising income tax rates or social security contribution rates increase the production wage, to which employers react by reducing their employment demand. Labour supply is modelled via the share of the labour force of the two age groups in the total population. These equations have also been estimated as Tobit models, with the restrictions of being positive but below 0.9 and 0.5 respectively. Labour supply depends positively on the real net wage and, as employment, negatively on the wedge between the gross and the net wage. In the wage-price system, gross wages, the consumer price index CPI (to be precise, the harmonised index of consumer prices HICP for Slovenia), and various deflators are determined. The gross wage rate depends on the price level, labour productivity and the unemployment rate. This equation is based on a bargaining model of the labour market, where the relative bargaining power of the employees (or the trade unions) is negatively affected by unemployment. The consumer price index is linked to the private consumption deflator. The latter depends on domestic and international factors. Domestic cost factors comprise unit labour costs and the capacity utilisation rate. The inclusion of the capacity utilisation rate in the price equation represents a channel for closing an output gap by increasing prices in the case of over-utilisation of capacities and by decreasing prices if actual production falls behind potential GDP . Foreign influences on Slovenian consumer prices are approximated by the import deflator. The public consumption deflator is linked to the most important cost factor of the public sector, which is public consumption. Public consumption includes purchases of goods and services and the wage costs of public employees. Similarly to consumer prices, both the investment and the export deflators are influenced by domestic and imported cost elements. The former are approximated by the unit labour costs while the latter are captured by the import deflator. Finally, the import deflator is influenced by the oil price in euro as a proxy for international raw material prices, which constitute an important determinant of the price level in a small open economy like Slovenia. On the money market, the short-term interest rate is linked to its Euro Area counterpart so as to capture Slovenia’s Euro Area membership and the resulting gradual adjustment of interest rates in Slovenia towards the Euro Area average. In the same vein, the long-term Euro Area interest rate is included in the equation determining the long-term interest rate in Slovenia. In addition, the long-term interest rate is linked to the short-term rate, representing the term structure of interest rates. Furthermore, the long-term interest rate is influenced by the debt to GDP ratio, representing a risk premium that rises with 275 K. WEYERSTRASS, R. NECK, D. BLUESCHKE, B. MAJCEN, A. SRAKAR, M. VERBIČ | SLOPOL10 ... the debt ratio. The foreign exchange market is modelled by the real effective exchange rate against a group of 41 countries. Due to Slovenia’s membership of the Euro Area, the nominal exchange rate is exogenous for Slovenia. However, the real exchange rate is still endogenous, even for the Euro Area countries, since it also depends on domestic price developments. Furthermore, the real effective exchange rate is an important determinant of exports and imports. When determining the effective exchange rate for Slovenia, it has to be taken into account that the country has only been a Euro Area member state since 2007. As the time series on which the estimations of the behavioural equations are based include the period before Slovenia ’ s Euro Area accession in 2007, the bilateral exchange rate between the Slovenian tolar and the euro is included as one of the explanatory variables in the real effective exchange rate equation. In addition, the exchange rate between the euro and the US dollar is considered. Furthermore, inflation in Slovenia is a regressor. To be theoretically consistent, the inflation differential between Slovenia and the group of countries forming the base for the real effective exchange rate should have been taken. However, this would have involved information about price developments in 41 countries, and for these exogenous variables assumptions had to be made for ex-post simulations. In the government sector of the model, the most important expenditure and revenue items of the Slovenian budget are determined. Social security contributions by employees are calculated by multiplying the average social security contribution rate by the gross wage rate and the number of employees. In the same vein, income tax payments by employees are determined by multiplying the average income tax rate by the gross wage rate and the number of employees. In a behavioural equation, social security payments by companies are linked to social security contributions by employees. Profit tax payments by companies are explained by GDP as an indicator for the economic situation, taking account of the fact that profits and hence profit tax payments display a strongly pro-cyclical behaviour. V alue added tax revenues depend on the value added tax rate and on private consumption. Other direct and indirect taxes are determined via their relation to nominal GDP , which is exogenous and has to be extrapolated in ex ante simulations, as for all other exogenous variables. Interest payments on public debt depend on the lagged debt level and on the long-term interest rate. Public consumption and transfer payments to private households as well as the remaining public expenditures and revenues are exogenous. By definition, the budget balance is given by the difference between total government revenues and expenditures. The public debt level is extrapolated using the budget balance equation. The model is closed by a number of identities and definition equations. 3. TESTS FOR STATIONARITY OF THE TIME SERIES As can be seen from Table A2 in the Appendix, it turns out that most level variables are I(1). Only a few variables are stationary in levels. These are the output gap (be construction, this variable should be stationary), the real interest rate, the real GDP growth rate, the labour force and employment of older people (very small numbers), the user cost of capital, and changes in inventories (as expected). For the budget balance in relation to GDP, the stationarity tests are inconclusive, although in the longer term this 276 ECONOMIC AND BUSINESS REVIEW | VOL. 20 | No. 2 | 2018 variable should be stationary. Also for the average real gross and net wage, the stationarity results are inconclusive, although one would expect these variables to increase over time. However, according to the data in our database, the average real wage per employee declined between 1996 and 2003, then rose until 2011, before decreasing again somewhat. We also tested for cointegration between those time series where we suspected long- run relations to hold. In those cases where cointegration seemed to be present, we used error-correction models as dynamic specifications for these relations while estimations in levels or first differences were tried when tests indicated the absence of long-run relations between stationary or between I(1) variables. The tests support our suspicion of cointegration between the variables we included in the behavioural equations. The detailed results can be found in Table A3 in the Appendix. 4. MODEL EQUATIONS In this section, the model equations are listed in detail, starting with the behavioural equations and then presenting the model identities. 4.1. Behavioural Equations Hereinafter, R² is the adjusted coefficient of determination, BG(p) is the Breusch-Godfrey Lagrange Multiplier statistic, a test for serial correlation up to lag p; * , **, *** denote rejection of the null hypothesis of no serial correlation at the 10, 5, 1 percent significance level respectively; t-statistics are given in parentheses below coefficients. 7 people (very small numbers), the user cost of capital, and changes in inventories (as expected). For the budget balance in relation to GDP, the stationarity tests are inconclusive, although in the longer term this variable should be stationary. Also for the average real gross and net wage, the stationarity results are inconclusive, although one would expect these variables to increase over time. However, according to the data in our database, the average real wage per employee declined between 1996 and 2003, then rose until 2011, before decreasing again somewhat. We also tested for cointegration between those time series where we suspected long-run relations to hold. In those cases where cointegration seemed to be present, we used error-correction models as dynamic specifications for these relations while estimations in levels or first differences were tried when tests indicated the absence of long-run relations between stationary or between I(1) variables. The tests support our suspicion of cointegration between the variables we included in the behavioural equations. The detailed results can be found in Table A3 in the Appendix. 4. Model Equations In this section, the model equations are listed in detail, starting with the behavioural equations and then presenting the model identities. 4.1. Behavioural Equations Hereinafter, R ² is the adjusted coefficient of determination, BG(p) is the Breusch-Godfrey Lagrange Multiplier statistic, a test for serial correlation up to lag p; * , **, *** denote rejection of the null hypothesis of no serial correlation at the 10, 5, 1 percent significance level respectively; t-statistics are given in parentheses below coefficients. Trend TFP LOG(TRENDTFP) = –4.588302 + 0.009127 * LOG(GERDR(–1)) + 0.384806 * LOG(LFTERSHARE) (–145.3956) (3.105505) (28.58483) + 0.309750 * LOG(INVR/GDPR) (15.03015) Adj. R ² = 0.923320 F-stat = 318.0849 BG(2) = 40.364*** Private Consumption LOG(CR/CR(–4)) = 0.321936 + 0.282529 * LOG(INCOMER/INCOMER(–4)) (1.108405) (5.481512) – 0.121486 * LOG(CR(–4)) + 0.081661 * LOG(INCOMER(–4)) (–7.369967) (2.362665) – 0.006417 * GOV10YR – 0.062606 D2013q1 (–5.068519) (–3.531924) Adj. R ² = 0.612852 F-stat = 24.74484 BG(2) = 6.503145** 277 K. WEYERSTRASS, R. NECK, D. BLUESCHKE, B. MAJCEN, A. SRAKAR, M. VERBIČ | SLOPOL10 ... 8 Private Gross Fixed Capital Formation LOG(PRINVR/PRINVR(–4)) = –0.000824 + 0.542725 * LOG(PRINVR(–1)/PRINVR(–5)) (–0.106209) (6.891356) + 0.404963 * LOG(INCOMER/INCOMER(–4)) (2.163258) – 0.018054 * (UCC(–1) – UCC(–5)) – 0.163850 * D2010q3 (–4.114459) (–2.41256) – 0.141658 * D2014q4 (–2.174659) Adj. R ² = 0.672624 F-stat = 29.76431 BG(2) = 3.772958 Exports LOG(EXR/EXR(–4)) = 0. 549852+ 0.277227 * LOG(EXR(–1)/EXR(–5)) (4.119548) (5.136417) + 0. 815406* LOG(WTRADE/WTRADE(–4)) (13.78450) – 0.321950* LOG(REER(–4)/REER(–8)) – 0.287643 * LOG(EXR(–4)) (–3.401803) (–4.888083) + 0.411336 * LOG(WTRADE(–4)) + 0.033620 D2007 – 0.026177 (D2013+D2013) (4.991134) (2.831993) (–2.808663) Adj. R ² = 0.917547 F-stat = 120.2305 BG(2) = 3.249562 Imports LOG(IMPR/IMPR(–4)) = –5.038052 + 1.315281 * LOG(DEMAND(–1)/DEMAND(–5)) (–3. 231196) (9.747473) + 0.801468* LOG(REER(–2)/REER(–6)) (2.011144) – 0.831232* LOG(REER(–3)/REER(–7)) – 0.480082 * LOG(IMPR(–4)) (–2.024690) (–2.652671) + 0.649493 * LOG(DEMAND(–4)) + 0.642609 * LOG(REER(–4)) (2.294327) (1.909966) + 0.090691 * D1998q1 – 0.200624 * D2009q1 (1.739119) (–4.110804) Adj. R ² = 0.684522 F-stat = 21.61303 BG(2) = 1.195105 9 Employment 15 to 64 EMP1564/POP1564 = –0.617752 + 0.473440 * EMP1564(–4)/POP1564(–4) + 0.200109 * LOG(GDPR) (–3.013194) (5.660659) (7.137335) – 0.044223 * LOG(NETWAGER) – 0.071028 * LOG(WEDGE) (–1.931810) (–5.892452) Employment 65+ EMP65PLUS/POP65PLUS = –0.088596 + 0.601889 * EMP65PLUS(-1)/POP65PLUS(-1) (–0.684680) (6.271412) + 0.057105 * LOG(GDPR) – 0.048881 * LOG(NETWAGEN+WEDGE) (1.928939) (–2.436480) Labour Supply 15 to 64 LF1564/POP1564 = 0.216732 + 0.694325 * LF1564(-4)/POP1564(-4) (4.602100) (10.31312) + 0.145252 * LOG(NETWAGER/NETWAGER(–4)) (4.829452) Labour Supply 65+ LF65PLUS/POP65PLUS = –0.170715+ 0.380958 * LF65PLUS(–1)/POP65PLUS(–1) (–1.207595) (3.843020) + 0.036490 * LOG(NETWAGER) – 0.018406 D2015 (2.213463) (–3.537480) – 0.010935 * LOG(WEDGE) – 0.011630 * (D2012+D2013) (–2.216665) (–2.812858) Average Gross Wage LOG(AGWN/AGWN(–4)) = 0.238652 + 0.599927 * LOG(AGWN(–1)/AGWN(–5)) (2.517697) (7.324412) + 0.133776 * LOG(CPI/CPI(–4)) + 0.114755 * LOG(PROD/PROD(–4)) (2.223294) (2.480250) – 0.003440 * UR – 0.055291 * LOG(AGWN(–4)/CPI(–4)) (–2.503514) (–2.175832) – 0.030158 * D2012q2 (–2.402247) Adj. R ² = 0.828677 F-stat = 61.46166 BG(2) = 2.439687 278 ECONOMIC AND BUSINESS REVIEW | VOL. 20 | No. 2 | 2018 9 Employment 15 to 64 EMP1564/POP1564 = –0.617752 + 0.473440 * EMP1564(–4)/POP1564(–4) + 0.200109 * LOG(GDPR) (–3.013194) (5.660659) (7.137335) – 0.044223 * LOG(NETWAGER) – 0.071028 * LOG(WEDGE) (–1.931810) (–5.892452) Employment 65+ EMP65PLUS/POP65PLUS = –0.088596 + 0.601889 * EMP65PLUS(-1)/POP65PLUS(-1) (–0.684680) (6.271412) + 0.057105 * LOG(GDPR) – 0.048881 * LOG(NETWAGEN+WEDGE) (1.928939) (–2.436480) Labour Supply 15 to 64 LF1564/POP1564 = 0.216732 + 0.694325 * LF1564(-4)/POP1564(-4) (4.602100) (10.31312) + 0.145252 * LOG(NETWAGER/NETWAGER(–4)) (4.829452) Labour Supply 65+ LF65PLUS/POP65PLUS = –0.170715+ 0.380958 * LF65PLUS(–1)/POP65PLUS(–1) (–1.207595) (3.843020) + 0.036490 * LOG(NETWAGER) – 0.018406 D2015 (2.213463) (–3.537480) – 0.010935 * LOG(WEDGE) – 0.011630 * (D2012+D2013) (–2.216665) (–2.812858) Average Gross Wage LOG(AGWN/AGWN(–4)) = 0.238652 + 0.599927 * LOG(AGWN(–1)/AGWN(–5)) (2.517697) (7.324412) + 0.133776 * LOG(CPI/CPI(–4)) + 0.114755 * LOG(PROD/PROD(–4)) (2.223294) (2.480250) – 0.003440 * UR – 0.055291 * LOG(AGWN(–4)/CPI(–4)) (–2.503514) (–2.175832) – 0.030158 * D2012q2 (–2.402247) Adj. R ² = 0.828677 F-stat = 61.46166 BG(2) = 2.439687 10 CPI LOG(CPI/CPI(–4)) = –0.000764 + 0.860254 * LOG(CPI(–1)/CPI(–5)) (–0.520422) (16.41307) + 0.119368 * LOG(CDEF/CDEF(–4)) (2.347029) – 0.024320 * LOG(CPI(–4))-LOG(CDEF(–4)) – 0.024477 * D2008q4 (–2.247985) (–3.425420) Adj. R ² = 0.942442 F-stat = 303.9159 BG(2) = 7.259309** Private Consumption Deflator LOG(CDEF/CDEF(–4)) = –0. 635911+ 0.270101* LOG(AGWN/AGWN(–4)) (–2.801746) (2.994393) + 0.129630* LOG(IMPDEF(–6)/IMPDEF(–10)) (2.534036) – 0.268560 * LOG(CDEF(–4)) + 0.101022 * LOG(AGWN(–4)) (–3.637782) (3.249838) + 0.133540 * LOG(UTIL(–1)) + 0.091529 * LOG(IMPDEF(–4)) (2.641737) (1.854469) Adj. R ² = 0.571235 F-stat = 17.20944 BG(2) = 16.17359*** Public Consumption Deflator LOG(GDEF/GDEF(–4)) = 0.119450 + 0.544327 * LOG(GDEF(–1)/GDEF(–5)) (1.851414) (6.264521) + 0.090745 * LOG(GNFIN/GNFIN(–4)) – 0.086096 * LOG(GDEF(–4)) (2.283731) (–3.041525) + 0.038165 * LOG(GNFIN(–4)) (3.062869) Adj. R ² = 0.680608 F-stat = 42.55355 BG(2) = 1.793151 Investment Deflator LOG(INVDEF/INVDEF(–4)) = 0.010428 + 0.216076 * LOG(ULC/ULC(–4)) (5.262049) (4.098676) + 0.141856 * LOG(IMPDEF/IMPDEF(–4)) (2.601534) + 0.042883 * D1997q1 + 0.046206 * D1998q4 (2.655108) (2.855100) – 0.052778 * D2000q4 (–3.160315) Adj. R ² = 0.342428 F-stat = 9.227795 BG(2) = 31.20401 279 K. WEYERSTRASS, R. NECK, D. BLUESCHKE, B. MAJCEN, A. SRAKAR, M. VERBIČ | SLOPOL10 ... 10 CPI LOG(CPI/CPI(–4)) = –0.000764 + 0.860254 * LOG(CPI(–1)/CPI(–5)) (–0.520422) (16.41307) + 0.119368 * LOG(CDEF/CDEF(–4)) (2.347029) – 0.024320 * LOG(CPI(–4))-LOG(CDEF(–4)) – 0.024477 * D2008q4 (–2.247985) (–3.425420) Adj. R ² = 0.942442 F-stat = 303.9159 BG(2) = 7.259309** Private Consumption Deflator LOG(CDEF/CDEF(–4)) = –0. 635911+ 0.270101* LOG(AGWN/AGWN(–4)) (–2.801746) (2.994393) + 0.129630* LOG(IMPDEF(–6)/IMPDEF(–10)) (2.534036) – 0.268560 * LOG(CDEF(–4)) + 0.101022 * LOG(AGWN(–4)) (–3.637782) (3.249838) + 0.133540 * LOG(UTIL(–1)) + 0.091529 * LOG(IMPDEF(–4)) (2.641737) (1.854469) Adj. R ² = 0.571235 F-stat = 17.20944 BG(2) = 16.17359*** Public Consumption Deflator LOG(GDEF/GDEF(–4)) = 0.119450 + 0.544327 * LOG(GDEF(–1)/GDEF(–5)) (1.851414) (6.264521) + 0.090745 * LOG(GNFIN/GNFIN(–4)) – 0.086096 * LOG(GDEF(–4)) (2.283731) (–3.041525) + 0.038165 * LOG(GNFIN(–4)) (3.062869) Adj. R ² = 0.680608 F-stat = 42.55355 BG(2) = 1.793151 Investment Deflator LOG(INVDEF/INVDEF(–4)) = 0.010428 + 0.216076 * LOG(ULC/ULC(–4)) (5.262049) (4.098676) + 0.141856 * LOG(IMPDEF/IMPDEF(–4)) (2.601534) + 0.042883 * D1997q1 + 0.046206 * D1998q4 (2.655108) (2.855100) – 0.052778 * D2000q4 (–3.160315) Adj. R ² = 0.342428 F-stat = 9.227795 BG(2) = 31.20401 11 Export Deflator LOG(EXPDEF/EXPDEF(–4)) = 0.691182 + 0.477104 * LOG(IMPDEF/IMPDEF(–4)) (5.368551) (13.53162) – 0.636126 * LOG(EXPDEF(–4)) + 0.403268 * LOG(IMPDEF(–4)) (-6.693435) (6.843747) + 0.046780 LOG(AGWN(–4)) (3.329078) Adj. R ² = 0.785893 F-stat = 73.49374 BG(2) = 10.24065*** Import Deflator LOG(IMPDEF/IMPDEF(–4)) = 1.688217 + 0.064189 * LOG(OILEUR/OILEUR(–4)) (6.514300) (8.883464) – 0.427363 * LOG(IMPDEF(–4)) + 0.070433 * LOG(OILEUR(–4)) (–6.675438) (7.561347) – 0.040262 * D2009 + 0.028375 * D2010 (–3.950683) (2.861353) Adj. R ² = 0.698642 F-stat = 37.62936 BG(2) = 28.40523*** Short-term Interest Rate SITBOR3M–SITBOR3M(–4) = 0.072921 + 0.583728 * (SITBOR3M(–1) –SITBOR3M(–5)) (1.110144) (10.69963) + 0.510182 * (EUR3M–EUR3M(–4)) (7.271125) – 0.453068 * (SITBOR3M(–4) –EUR3M(–4)) (–6.395199) Adj. R ² = 0.859096 F-stat = 159.5222 BG(2) = 23.92325*** Long-term Interest Rate GOV10Y–GOV10Y(–4) = –0.116529 + 0.218874 * (SITBOR3M–SITBOR3M(–4)) (–0.780286) (2.522239) + 2.021775 * (EUR10Y–EUR10Y(–4)) (10.71268) + 1.694831 * LOG(DEBTGDP/DEBTGDP(-4)) – 1.856888 * D2004 (1.704599) (–3.693687) + 1.992136 * D2012 + 1.624226 * D2013 (4.029161) (3.083994) Adj. R ² = 0.679935 F-stat = 23.30579 BG(2) = 17.72585*** 280 ECONOMIC AND BUSINESS REVIEW | VOL. 20 | No. 2 | 2018 11 Export Deflator LOG(EXPDEF/EXPDEF(–4)) = 0.691182 + 0.477104 * LOG(IMPDEF/IMPDEF(–4)) (5.368551) (13.53162) – 0.636126 * LOG(EXPDEF(–4)) + 0.403268 * LOG(IMPDEF(–4)) (-6.693435) (6.843747) + 0.046780 LOG(AGWN(–4)) (3.329078) Adj. R ² = 0.785893 F-stat = 73.49374 BG(2) = 10.24065*** Import Deflator LOG(IMPDEF/IMPDEF(–4)) = 1.688217 + 0.064189 * LOG(OILEUR/OILEUR(–4)) (6.514300) (8.883464) – 0.427363 * LOG(IMPDEF(–4)) + 0.070433 * LOG(OILEUR(–4)) (–6.675438) (7.561347) – 0.040262 * D2009 + 0.028375 * D2010 (–3.950683) (2.861353) Adj. R ² = 0.698642 F-stat = 37.62936 BG(2) = 28.40523*** Short-term Interest Rate SITBOR3M–SITBOR3M(–4) = 0.072921 + 0.583728 * (SITBOR3M(–1) –SITBOR3M(–5)) (1.110144) (10.69963) + 0.510182 * (EUR3M–EUR3M(–4)) (7.271125) – 0.453068 * (SITBOR3M(–4) –EUR3M(–4)) (–6.395199) Adj. R ² = 0.859096 F-stat = 159.5222 BG(2) = 23.92325*** Long-term Interest Rate GOV10Y–GOV10Y(–4) = –0.116529 + 0.218874 * (SITBOR3M–SITBOR3M(–4)) (–0.780286) (2.522239) + 2.021775 * (EUR10Y–EUR10Y(–4)) (10.71268) + 1.694831 * LOG(DEBTGDP/DEBTGDP(-4)) – 1.856888 * D2004 (1.704599) (–3.693687) + 1.992136 * D2012 + 1.624226 * D2013 (4.029161) (3.083994) Adj. R ² = 0.679935 F-stat = 23.30579 BG(2) = 17.72585*** 12 Real Effective Exchange Rate LOG(REER/REER(–4)) = –0.007941 + 0.084268 * LOG(EURUSD/EURUSD(–4)) (–2.789133) (4.503065) + 0.280321 * LOG(SITEUR/SITEUR(–4)) (4.729566) + 0.678165 * LOG(GDPDEF/GDPDEF(–4)) + 0.037226 * D1998 (6.623438) (4.447943) + 0.031405 * D1999 (3.946994) Adj. R ² = 0.701605 F-stat = 38.14987 BG(2) = 31.90596*** Employers’ Social Security Contributions LOG(SOCCOMP/SOCCOMP(–4)) = –0.418600 + 0.941308 * LOG(SOCEMP/SOCEMP(–4)) (–7.290584) (14.45902) – 0.646844 * LOG(SOCCOMP(–4)) (–17.69022) + 0.682561 * LOG(SOCEMP(–4)) (19.67186) Adj. R ² = 0.888454 F-stat = 210.7419 BG(2) = 3.277950 Corporate Income Tax Payments INCTAXCORP–INCTAXCORP(–4) = –1717.275 + 1168.325 * LOG(GDPR/GDPR(–4)) (–3.778722) (5.918436) – 0.341519 * INCTAXCORP(–4) + 193.6532 * LOG(GDPR(–4)) (–4.077339) (3.780993) Adj. R ² = 0.421035 F-stat = 20.15009 BG(2) = 0.591128 Value Added Tax Revenues LOG(VAT) = –5.491826 + 1.054549 * LOG(CN) + 1.054032 * LOG(VATAXRATE) (–7.238066) (19.42491) (4.267224) – 0.336750 * D2000q1 – 0.630827 D2001q1 – 0.926044 D2002q1 (–2.658629) (–4.981327) (–7.337844) Adj. R ² = 0.883668 F-stat = 127.0950 BG(2) = 4.614928* 281 K. WEYERSTRASS, R. NECK, D. BLUESCHKE, B. MAJCEN, A. SRAKAR, M. VERBIČ | SLOPOL10 ... 12 Real Effective Exchange Rate LOG(REER/REER(–4)) = –0.007941 + 0.084268 * LOG(EURUSD/EURUSD(–4)) (–2.789133) (4.503065) + 0.280321 * LOG(SITEUR/SITEUR(–4)) (4.729566) + 0.678165 * LOG(GDPDEF/GDPDEF(–4)) + 0.037226 * D1998 (6.623438) (4.447943) + 0.031405 * D1999 (3.946994) Adj. R ² = 0.701605 F-stat = 38.14987 BG(2) = 31.90596*** Employers’ Social Security Contributions LOG(SOCCOMP/SOCCOMP(–4)) = –0.418600 + 0.941308 * LOG(SOCEMP/SOCEMP(–4)) (–7.290584) (14.45902) – 0.646844 * LOG(SOCCOMP(–4)) (–17.69022) + 0.682561 * LOG(SOCEMP(–4)) (19.67186) Adj. R ² = 0.888454 F-stat = 210.7419 BG(2) = 3.277950 Corporate Income Tax Payments INCTAXCORP–INCTAXCORP(–4) = –1717.275 + 1168.325 * LOG(GDPR/GDPR(–4)) (–3.778722) (5.918436) – 0.341519 * INCTAXCORP(–4) + 193.6532 * LOG(GDPR(–4)) (–4.077339) (3.780993) Adj. R ² = 0.421035 F-stat = 20.15009 BG(2) = 0.591128 Value Added Tax Revenues LOG(VAT) = –5.491826 + 1.054549 * LOG(CN) + 1.054032 * LOG(VATAXRATE) (–7.238066) (19.42491) (4.267224) – 0.336750 * D2000q1 – 0.630827 D2001q1 – 0.926044 D2002q1 (–2.658629) (–4.981327) (–7.337844) Adj. R ² = 0.883668 F-stat = 127.0950 BG(2) = 4.614928* 13 Interest Payments on Public Debt LOG(INTEREST) = –1.966945+ 0.832199* LOG(INTEREST(–4)) (–1.894332) (17.18193) + 0.242440 * LOG(DEBT(–4)*GOV10Y) (2.378300) + 1.454346 * (D2010q2+D2010q3) + 0.2866858 * q1 (5.976520) (3.071885) Adj. R ² = 0.859831 F-stat = 122.1512 BG(2) = 1.288664 4.2. Identities AGWR = AGWN / CPI * 100 BALANCE = TGRN – TGEN BALANCEGDP = BALANCE / GDPN * 100 CAGDP = CAN / GDPN * 100 CAN = EXR * EXPDEF / 100 – IMPR * IMPDEF / 100 CAPR = (1 – DEPR / 100) * CAPR(–1) + INVR CN = CR * CDEF / 100 DEBT = DEBT(–1) – BALANCE + BANKCAP + DEBTADJ DEBTGDP = DEBT / (GDPN + GDPN(–1) + GDPN(–2) + GDPN(–3)) * 100 DEMAND = INVR + CR + GR + EXR EMP = EMP1564 + EMP65PLUS GAP = (GDPR – YPOT) / YPOT * 100 GDPDEF = GDPN / GDPR * 100 GDPN = CN + GN + (INVR + INVENTR) * INVDEF / 100 + CAN GDPR = CR + GR + INVR + INVENTR + EXR – IMPR GERDR = GERD / INVDEF * 100 GINVR = GINVN / INVDEF * 100 GN = GNFIN + GN_REST GOV10YR = GOV10Y – INFL GR = GN / GDEF * 100 GRGDPR = GDPR / GDPR(–4) * 100 – 100 GRYPOT = (YPOT / YPOT(–4) – 1) * 100 INCOME = GDPN+TRANSFERSN–SOCTOTAL–INCTAX–VAT–TAXDIRREST–TAXINDIRREST INCOMER = INCOME / CPI * 100 INCTAX = INCTAXPERS + INCTAXCORP INCTAXPERS = INCTAXRATE * (AGWN * EMP / 1000) / 1000 INFL = (CPI / CPI(–4) – 1) * 100 INVN = INVR * INVDEF / 100 INVR = PRINVR + GINVR + GERDR LF = LF1564 + LF65PLUS LOG(YPOT) = 0.65 * LOG(TRENDEMP) + (1 - 0.65) * LOG(CAPR) + LOG(TRENDTFP) NETWAGEN = AGWN – WEDGE NETWAGER = NETWAGEN / CPI * 100 OILEUR = OIL / EURUSD PRIMBALANCE = BALANCE + INTEREST 282 ECONOMIC AND BUSINESS REVIEW | VOL. 20 | No. 2 | 2018 4.2. Identities AGWR = AGWN / CPI * 100 BALANCE = TGRN – TGEN BALANCEGDP = BALANCE / GDPN * 100 CAGDP = CAN / GDPN * 100 CAN = EXR * EXPDEF / 100 – IMPR * IMPDEF / 100 CAPR = (1 – DEPR / 100) * CAPR(–1) + INVR CN = CR * CDEF / 100 DEBT = DEBT(–1) – BALANCE + BANKCAP + DEBTADJ DEBTGDP = DEBT / (GDPN + GDPN(–1) + GDPN(–2) + GDPN(–3)) * 100 DEMAND = INVR + CR + GR + EXR EMP = EMP1564 + EMP65PLUS GAP = (GDPR – YPOT) / YPOT * 100 GDPDEF = GDPN / GDPR * 100 GDPN = CN + GN + (INVR + INVENTR) * INVDEF / 100 + CAN GDPR = CR + GR + INVR + INVENTR + EXR – IMPR GERDR = GERD / INVDEF * 100 GINVR = GINVN / INVDEF * 100 GN = GNFIN + GN_REST GOV10YR = GOV10Y – INFL GR = GN / GDEF * 100 GRGDPR = GDPR / GDPR(–4) * 100 – 100 GRYPOT = (YPOT / YPOT(–4) – 1) * 100 INCOME = GDPN+TRANSFERSN–SOCTOTAL–INCTAX–VAT– TAXDIRREST–TAXINDIRREST INCOMER = INCOME / CPI * 100 INCTAX = INCTAXPERS + INCTAXCORP INCTAXPERS = INCTAXRATE * (AGWN * EMP / 1000) / 1000 INFL = (CPI / CPI(–4) – 1) * 100 INVN = INVR * INVDEF / 100 INVR = PRINVR + GINVR + GERDR LF = LF1564 + LF65PLUS LOG(YPOT) = 0.65 * LOG(TRENDEMP) + (1 - 0.65) * LOG(CAPR) + LOG(TRENDTFP) NETWAGEN = AGWN – WEDGE NETWAGER = NETWAGEN / CPI * 100 OILEUR = OIL / EURUSD PRIMBALANCE = BALANCE + INTEREST PRIMBALANCEGDP = PRIMBALANCE / GDPN * 100 PROD = GDPR / EMP * 100 SOCEMP = SOCEMPRATE * (AGWN * EMP / 1000) / 1000 SOCTOTAL = SOCCOMP + SOCEMP TAXDIRREST = TAXDIRRATE * GDPN / 100 TAXINDIRREST = TAXINDIRRATE * GDPN / 100 14 PRIMBALANCEGDP = PRIMBALANCE / GDPN * 100 PROD = GDPR / EMP * 100 SOCEMP = SOCEMPRATE * (AGWN * EMP / 1000) / 1000 SOCTOTAL = SOCCOMP + SOCEMP TAXDIRREST = TAXDIRRATE * GDPN / 100 TAXINDIRREST = TAXINDIRRATE * GDPN / 100 TGEN = GNFIN + GINVN + TRANSFERSN + INTEREST + EXPREST TGRN = VAT + SOCTOTAL + INCTAX + TAXDIRREST + TAXINDIRREST + REVREST TRENDEMP = LF * (1 – NAIRU_EU / 100) UCC = GOV10YR + DEPR ULC = AGWN / PROD UN = LF – EMP UN1564 = LF1564 – EMP1564 UR = UN / LF * 100 UR1564 = UN1564 / LF1564 * 100 UTIL = GDPR / YPOT * 100 WEDGE = AGWN * (INCTAXRATE + SOCEMPRATE) 5. Ex-post Simulation Figures A1–A12 in the Appendix show the results of a dynamic ex-post simulation of the model over the period 1999 to 2015 for the key macroeconomic variables. In addition to the visual inspection, we tested the quality of the ex-post forecasting performance of the model formally. As quality criteria we chose the root mean squared error (RMSE) or the root mean squared percent error (RMSPE), the mean absolute percent error (MAPE) or the mean absolute error (MAE), and Theil’s inequality coefficient (THEIL). Regarding the Theil coefficient, we chose the U2 coefficient, defined by the following formula: THEIL = 𝐹𝐹 ! − 𝐴𝐴 ! ² ! ! ! ! 𝐴𝐴 ! ² ! ! ! ! where A i and 𝐹𝐹 ! denote the actual realisations and forecasts of changes in the underlying variables. The benchmark is the no-change forecast. In this case, THEIL will take the value 1. Values below 1 show an improvement over the simple no-change forecast (Theil 1966). The RMSE, the RMSPE, the MAE and the MAPE are defined as follows (Shcherbakov et al., 2013): RM SE = 1 𝑛𝑛 𝐹𝐹 ! − 𝐴𝐴 ! ² ! ! ! ! RM SPE = 1 𝑛𝑛 100 ∗ 𝐹𝐹 ! − 𝐴𝐴 ! 𝐴𝐴 ! ! ! ! ! ! 14 PRIMBALANCEGDP = PRIMBALANCE / GDPN * 100 PROD = GDPR / EMP * 100 SOCEMP = SOCEMPRATE * (AGWN * EMP / 1000) / 1000 SOCTOTAL = SOCCOMP + SOCEMP TAXDIRREST = TAXDIRRATE * GDPN / 100 TAXINDIRREST = TAXINDIRRATE * GDPN / 100 TGEN = GNFIN + GINVN + TRANSFERSN + INTEREST + EXPREST TGRN = VAT + SOCTOTAL + INCTAX + TAXDIRREST + TAXINDIRREST + REVREST TRENDEMP = LF * (1 – NAIRU_EU / 100) UCC = GOV10YR + DEPR ULC = AGWN / PROD UN = LF – EMP UN1564 = LF1564 – EMP1564 UR = UN / LF * 100 UR1564 = UN1564 / LF1564 * 100 UTIL = GDPR / YPOT * 100 WEDGE = AGWN * (INCTAXRATE + SOCEMPRATE) 5. Ex-post Simulation Figures A1–A12 in the Appendix show the results of a dynamic ex-post simulation of the model over the period 1999 to 2015 for the key macroeconomic variables. In addition to the visual inspection, we tested the quality of the ex-post forecasting performance of the model formally. As quality criteria we chose the root mean squared error (RMSE) or the root mean squared percent error (RMSPE), the mean absolute percent error (MAPE) or the mean absolute error (MAE), and Theil’s inequality coefficient (THEIL). Regarding the Theil coefficient, we chose the U2 coefficient, defined by the following formula: THEIL = 𝐹𝐹 ! − 𝐴𝐴 ! ² ! ! ! ! 𝐴𝐴 ! ² ! ! ! ! where A i and 𝐹𝐹 ! denote the actual realisations and forecasts of changes in the underlying variables. The benchmark is the no-change forecast. In this case, THEIL will take the value 1. Values below 1 show an improvement over the simple no-change forecast (Theil 1966). The RMSE, the RMSPE, the MAE and the MAPE are defined as follows (Shcherbakov et al., 2013): RM SE = 1 𝑛𝑛 𝐹𝐹 ! − 𝐴𝐴 ! ² ! ! ! ! RM SPE = 1 𝑛𝑛 100 ∗ 𝐹𝐹 ! − 𝐴𝐴 ! 𝐴𝐴 ! ! ! ! ! ! 283 K. WEYERSTRASS, R. NECK, D. BLUESCHKE, B. MAJCEN, A. SRAKAR, M. VERBIČ | SLOPOL10 ... TGEN = GNFIN + GINVN + TRANSFERSN + INTEREST + EXPREST TGRN = VAT + SOCTOTAL + INCTAX + TAXDIRREST + TAXINDIRREST + REVREST TRENDEMP = LF * (1 – NAIRU_EU / 100) UCC = GOV10YR + DEPR ULC = AGWN / PROD UN = LF – EMP UN1564 = LF1564 – EMP1564 UR = UN / LF * 100 UR1564 = UN1564 / LF1564 * 100 UTIL = GDPR / YPOT * 100 WEDGE = AGWN * (INCTAXRATE + SOCEMPRATE) 5. EX-POST SIMULATION Figures A1–A12 in the Appendix show the results of a dynamic ex-post simulation of the model over the period 1999 to 2015 for the key macroeconomic variables. In addition to the visual inspection, we tested the quality of the ex-post forecasting performance of the model formally. As quality criteria we chose the root mean squared error (RMSE) or the root mean squared percent error (RMSPE), the mean absolute percent error (MAPE) or the mean absolute error (MAE), and Theil’s inequality coefficient (THEIL). Regarding the Theil coefficient, we chose the U2 coefficient, defined by the following formula: where A i and F i denote the actual realisations and forecasts of changes in the underlying variables. The benchmark is the no-change forecast. In this case, THEIL will take the value 1. Values below 1 show an improvement over the simple no-change forecast (Theil 1966). The RMSE, the RMSPE, the MAE and the MAPE are defined as follows (Shcherbakov et al., 2013): 14 PRIMBALANCEGDP = PRIMBALANCE / GDPN * 100 PROD = GDPR / EMP * 100 SOCEMP = SOCEMPRATE * (AGWN * EMP / 1000) / 1000 SOCTOTAL = SOCCOMP + SOCEMP TAXDIRREST = TAXDIRRATE * GDPN / 100 TAXINDIRREST = TAXINDIRRATE * GDPN / 100 TGEN = GNFIN + GINVN + TRANSFERSN + INTEREST + EXPREST TGRN = VAT + SOCTOTAL + INCTAX + TAXDIRREST + TAXINDIRREST + REVREST TRENDEMP = LF * (1 – NAIRU_EU / 100) UCC = GOV10YR + DEPR ULC = AGWN / PROD UN = LF – EMP UN1564 = LF1564 – EMP1564 UR = UN / LF * 100 UR1564 = UN1564 / LF1564 * 100 UTIL = GDPR / YPOT * 100 WEDGE = AGWN * (INCTAXRATE + SOCEMPRATE) 5. Ex-post Simulation Figures A1–A12 in the Appendix show the results of a dynamic ex-post simulation of the model over the period 1999 to 2015 for the key macroeconomic variables. In addition to the visual inspection, we tested the quality of the ex-post forecasting performance of the model formally. As quality criteria we chose the root mean squared error (RMSE) or the root mean squared percent error (RMSPE), the mean absolute percent error (MAPE) or the mean absolute error (MAE), and Theil’s inequality coefficient (THEIL). Regarding the Theil coefficient, we chose the U2 coefficient, defined by the following formula: THEIL = 𝐹𝐹 ! − 𝐴𝐴 ! ² ! ! ! ! 𝐴𝐴 ! ² ! ! ! ! where A i and 𝐹𝐹 ! denote the actual realisations and forecasts of changes in the underlying variables. The benchmark is the no-change forecast. In this case, THEIL will take the value 1. Values below 1 show an improvement over the simple no-change forecast (Theil 1966). The RMSE, the RMSPE, the MAE and the MAPE are defined as follows (Shcherbakov et al., 2013): RM SE = 1 𝑛𝑛 𝐹𝐹 ! − 𝐴𝐴 ! ² ! ! ! ! RM SPE = 1 𝑛𝑛 100 ∗ 𝐹𝐹 ! − 𝐴𝐴 ! 𝐴𝐴 ! ! ! ! ! ! 14 PRIMBALANCEGDP = PRIMBALANCE / GDPN * 100 PROD = GDPR / EMP * 100 SOCEMP = SOCEMPRATE * (AGWN * EMP / 1000) / 1000 SOCTOTAL = SOCCOMP + SOCEMP TAXDIRREST = TAXDIRRATE * GDPN / 100 TAXINDIRREST = TAXINDIRRATE * GDPN / 100 TGEN = GNFIN + GINVN + TRANSFERSN + INTEREST + EXPREST TGRN = VAT + SOCTOTAL + INCTAX + TAXDIRREST + TAXINDIRREST + REVREST TRENDEMP = LF * (1 – NAIRU_EU / 100) UCC = GOV10YR + DEPR ULC = AGWN / PROD UN = LF – EMP UN1564 = LF1564 – EMP1564 UR = UN / LF * 100 UR1564 = UN1564 / LF1564 * 100 UTIL = GDPR / YPOT * 100 WEDGE = AGWN * (INCTAXRATE + SOCEMPRATE) 5. Ex-post Simulation Figures A1–A12 in the Appendix show the results of a dynamic ex-post simulation of the model over the period 1999 to 2015 for the key macroeconomic variables. In addition to the visual inspection, we tested the quality of the ex-post forecasting performance of the model formally. As quality criteria we chose the root mean squared error (RMSE) or the root mean squared percent error (RMSPE), the mean absolute percent error (MAPE) or the mean absolute error (MAE), and Theil’s inequality coefficient (THEIL). Regarding the Theil coefficient, we chose the U2 coefficient, defined by the following formula: THEIL = 𝐹𝐹 ! − 𝐴𝐴 ! ² ! ! ! ! 𝐴𝐴 ! ² ! ! ! ! where A i and 𝐹𝐹 ! denote the actual realisations and forecasts of changes in the underlying variables. The benchmark is the no-change forecast. In this case, THEIL will take the value 1. Values below 1 show an improvement over the simple no-change forecast (Theil 1966). The RMSE, the RMSPE, the MAE and the MAPE are defined as follows (Shcherbakov et al., 2013): RM SE = 1 𝑛𝑛 𝐹𝐹 ! − 𝐴𝐴 ! ² ! ! ! ! RM SPE = 1 𝑛𝑛 100 ∗ 𝐹𝐹 ! − 𝐴𝐴 ! 𝐴𝐴 ! ! ! ! ! ! 284 ECONOMIC AND BUSINESS REVIEW | VOL. 20 | No. 2 | 2018 We took the RMSE and the MAE for interest rates, ratios (net exports, budget balance and public debt in relation to GDP), growth rates, interest rates, the inflation rate and the unemployment rate, and the RMSPE and the MAPE for all other variables. The results of these tests ascertaining the quality of the ex-post simulation are shown in Table A4 in the Appendix. Overall, the results are quite promising. The high values of the error statistics for the budget balance and net exports can be explained by the fact that in some cases the simulation misses the correct sign, leading to large errors. Among the demand components, for investment and imports the model simulation is worse than for the other GDP components. Employment and unemployment are in general tracked satisfactorily, with the exception of the labour market indicators of the older people, which is due to the very small absolute numbers of these variables. 6. CONCLUDING REMARKS The SLOPOL10 model as described above was obtained after a series of steps, following the general-to-specific methodology initiated by David Hendry and associates (see, e.g., Hendry 1995). We also conducted simulations of the model (both static and dynamic) with historical values of (non-controllable and policy) exogenous variables over the period of estimation and found reasonable tracking quality for most variables with respect to trends and turning points. This encourages us to use the model for policy analysis. Among these, policy simulations for fiscal policy design and optimal control experiments for determining optimal budgetary policies will be prominent. ACKNOWLEDGMENT The authors gratefully acknowledge financial support from the Slovenian Research Agency ARRS (contract no. 630-31/2016-1) and the Austrian Science Foundation FWF (project no. I 2764-G27). Helpful suggestions from two anonymous referees are gratefully acknowledged. The usual caveat applies. 15 MAE = 1 𝑛𝑛 𝐹𝐹 ! − 𝐴𝐴 ! ! ! ! ! MAPE = 1 𝑛𝑛 100 ∗ 𝐹𝐹 ! − 𝐴𝐴 ! 𝐹𝐹 ! ! ! ! ! We took the RMSE and the MAE for interest rates, ratios (net exports, budget balance and public debt in relation to GDP), growth rates, interest rates, the inflation rate and the unemployment rate, and the RMSPE and the MAPE for all other variables. The results of these tests ascertaining the quality of the ex-post simulation are shown in Table A4 in the Appendix. Overall, the results are quite promising. The high values of the error statistics for the budget balance and net exports can be explained by the fact that in some cases the simulation misses the correct sign, leading to large errors. Among the demand components, for investment and imports the model simulation is worse than for the other GDP components. Employment and unemployment are in general tracked satisfactorily, with the exception of the labour market indicators of the older people, which is due to the very small absolute numbers of these variables. 6. Concluding Remarks The SLOPOL10 model as described above was obtained after a series of steps, following the general-to- specific methodology initiated by David Hendry and associates (see, e.g., Hendry 1995). We also conducted simulations of the model (both static and dynamic) with historical values of (non-controllable and policy) exogenous variables over the period of estimation and found reasonable tracking quality for most variables with respect to trends and turning points. This encourages us to use the model for policy analysis. Among these, policy simulations for fiscal policy design and optimal control experiments for determining optimal budgetary policies will be prominent. Acknowledgment The authors gratefully acknowledge financial support from the Slovenian Research Agency ARRS (contract no. 630-31/2016-1) and the Austrian Science Foundation FWF (project no. I 2764-G27). Helpful suggestions from two anonymous referees are gratefully acknowledged. The usual caveat applies. References Blueschke, D., Weyerstrass, K., Neck, R. (2016), How Should Slovenia Design Fiscal Policies in the Government Debt Crisis? Emerging Markets Finance and Trade 52, 1562–1573. Havik, K., Mc Morrow, K., Orlandi, F., Planas, C., Raciborski, R., Röger, W., Rossi, A., Thum-Thysen, A., Vandermeulen, V. (2014), The Production Function Methodology for Calculating Potential Growth Rates and Output Gaps. European Commission Economic Papers 535, Brussels. Hendry, D. (1995), Dynamic Econometrics. Oxford University Press, Oxford. Lucas, R. (1976), Econometric Policy Evaluation: A Critique. In: Brunner, K., Meltzer, A. (eds.), The Phillips Curve and Labor Markets, Carnegie-Rochester Conference Series on Public Policy 1, Elsevier, New York, 19–46. 285 K. WEYERSTRASS, R. NECK, D. BLUESCHKE, B. MAJCEN, A. SRAKAR, M. VERBIČ | SLOPOL10 ... REFERENCES Blueschke, D., Weyerstrass, K., Neck, R. (2016), How Should Slovenia Design Fiscal Policies in the Government Debt Crisis? Emerging Markets Finance and Trade 52, 1562–1573. Havik, K., Mc Morrow, K., Orlandi, F., Planas, C., Raciborski, R., Röger, W., Rossi, A., Thum-Thysen, A., Vandermeulen, V . (2014), The Production Function Methodology for Calculating Potential Growth Rates and Output Gaps. European Commission Economic Papers 535, Brussels. Hendry, D. (1995), Dynamic Econometrics. Oxford University Press, Oxford. Lucas, R. (1976), Econometric Policy Evaluation: A Critique. In: Brunner, K., Meltzer, A. (eds.), The Phillips Curve and Labor Markets, Carnegie-Rochester Conference Series on Public Policy 1, Elsevier, New Y ork, 19–46. Neck, R., Blueschke, D., Weyerstrass, K. (2011), Optimal Macroeconomic Policies in a Financial and Economic Crisis: A Case Study for Slovenia. Empirica 38, 435–459. Newey, W.K., West, K.D. (1994), Automatic Lag Length Selection in Covariance Matrix Estimation, Review of Economic Studies , 61, 631–653. Sargan, J.D. (1964), Wages and Prices in the United Kingdom. A Study in Econometric Methodology. In: Hart, P .E., Mills, G., Whitaker, J.K. (eds.), Econometric Analysis for National Economic Planning . Butterworth, London, 25-59. Sargent, T. (1981), Interpreting Economic Time Series. Journal of Political Economy 89, 213-248. Shcherbakov, M.V ., Brebels, A., Shcherbakova, N.L., Tyukov, A.P ., Janovsky, T.A., Kamaev, V .A. (2013), A Survey of Forecast Error Measures, World Applied Sciences Journal 24, 171-176. Theil, H. (1966), Applied Economic Forecasting . Rand McNally, Chicago. Verbič, M. (2005), Macroeconomic Analysis of Fundamental Relationships of the Slovenian Economy. Institute for Economic Research, Ljubljana. Verbič, M. (2006), A Quarterly Econometric Model of the Slovenian Economy. Economic and Business Review 8 (3), 215–244. Weyerstrass, K., Neck, R. (2007), SLOPOL6: A Macroeconometric Model for Slovenia. International Business and Economics Research Journal 6 (11), 81–94. 15 MAE = 1 𝑛𝑛 𝐹𝐹 ! − 𝐴𝐴 ! ! ! ! ! MAPE = 1 𝑛𝑛 100 ∗ 𝐹𝐹 ! − 𝐴𝐴 ! 𝐹𝐹 ! ! ! ! ! We took the RMSE and the MAE for interest rates, ratios (net exports, budget balance and public debt in relation to GDP), growth rates, interest rates, the inflation rate and the unemployment rate, and the RMSPE and the MAPE for all other variables. The results of these tests ascertaining the quality of the ex-post simulation are shown in Table A4 in the Appendix. Overall, the results are quite promising. The high values of the error statistics for the budget balance and net exports can be explained by the fact that in some cases the simulation misses the correct sign, leading to large errors. Among the demand components, for investment and imports the model simulation is worse than for the other GDP components. Employment and unemployment are in general tracked satisfactorily, with the exception of the labour market indicators of the older people, which is due to the very small absolute numbers of these variables. 6. Concluding Remarks The SLOPOL10 model as described above was obtained after a series of steps, following the general-to- specific methodology initiated by David Hendry and associates (see, e.g., Hendry 1995). We also conducted simulations of the model (both static and dynamic) with historical values of (non-controllable and policy) exogenous variables over the period of estimation and found reasonable tracking quality for most variables with respect to trends and turning points. This encourages us to use the model for policy analysis. Among these, policy simulations for fiscal policy design and optimal control experiments for determining optimal budgetary policies will be prominent. Acknowledgment The authors gratefully acknowledge financial support from the Slovenian Research Agency ARRS (contract no. 630-31/2016-1) and the Austrian Science Foundation FWF (project no. I 2764-G27). Helpful suggestions from two anonymous referees are gratefully acknowledged. The usual caveat applies. References Blueschke, D., Weyerstrass, K., Neck, R. (2016), How Should Slovenia Design Fiscal Policies in the Government Debt Crisis? Emerging Markets Finance and Trade 52, 1562–1573. Havik, K., Mc Morrow, K., Orlandi, F., Planas, C., Raciborski, R., Röger, W., Rossi, A., Thum-Thysen, A., Vandermeulen, V. (2014), The Production Function Methodology for Calculating Potential Growth Rates and Output Gaps. European Commission Economic Papers 535, Brussels. Hendry, D. (1995), Dynamic Econometrics. Oxford University Press, Oxford. Lucas, R. (1976), Econometric Policy Evaluation: A Critique. In: Brunner, K., Meltzer, A. (eds.), The Phillips Curve and Labor Markets, Carnegie-Rochester Conference Series on Public Policy 1, Elsevier, New York, 19–46. 286 ECONOMIC AND BUSINESS REVIEW | VOL. 20 | No. 2 | 2018 Appendix Table A1: List of Variables Endogenous Variables AGWN Average gross wage, euro per employee AGWR Average gross wage real BALANCE Budget balance BALANCEGDP Budget balance in relation to GDP CAGDP Current account balance in percent of GDP CAN Current account balance CAPR Real capital stock CDEF Private consumption deflator CN Private consumption, nominal CPI Consumer price index CR Private consumption, real DEBT Public debt stock DEBTGDP Debt level in relation to GDP DEMAND Final demand, real EMP Total number of employees EMP1564 Employment, 15 to 64 years EMP65PLUS Employment, 65 years or older EXPDEF Export deflator EXR Exports of goods and services, real GAP Output gap in percent of potential GDP GDEF Public consumption deflator GDPDEF GDP deflator GDPN Nominal GDP GDPR Real GDP GERDR Real government R&D expenditures GINVR Real government investment GN Public consumption, national accounts, nominal GOV10Y 10 year government bond yield GOV10YR Real government bond yield GR Public consumption, real GRGDPR Real GDP growth rate GRYPOT Growth rate of potential GDP IMPDEF Import deflator IMPR Imports of goods and services, real INCOME Disposable income of private households, nominal INCOMER Disposable income of private households, real INCTAX Total income tax revenues INCTAXCORP Corporate income tax revenues INCTAXPERS Personal income tax revenues INFL Inflation rate 287 K. WEYERSTRASS, R. NECK, D. BLUESCHKE, B. MAJCEN, A. SRAKAR, M. VERBIČ | SLOPOL10 ... INTEREST Interest payments on public debt INVDEF Investment deflator INVN Gross fixed capital formation, nominal INVR Gross fixed capital formation, real LF Total labour force LF1564 Labour force, 15 to 64 years LF65PLUS Labour force, 65 years or older NETWAGEN Net wage, nominal NETWAGER Average net wage, real OILEUR Oil price in euro PRIMBALANCE Primary budget balance PRIMBALANCEGDP Primary budget balance in relation to GDP PRINVR Real private investment PROD Labour productivity REER Real effective exchange rate (deflator: consumer price indices, 42 trading partners) SITBOR3M 3 month interest rate before 2007, EURIBOR from 2007 onwards SOCCOMP Social security contributions by employers SOCEMP Social security contributions by employees SOCTOTAL Total social security contributions TAXDIRECT Other direct taxes TAXINDIRECT Other indirect taxes TGEN Total government expenditures TGRN Total government revenues TRENDEMP Trend of employment TRENDTFP Trend of total factor productivity UCC User cost of capital ULC Unit labour cost UN Total number of unemployed persons UN1564 Unemployment, 15 to 64 years UR Unemployment rate UR1564 Unemployment rate, 15 to 64 years UTIL Capacity utilisation rate VAT Value added tax revenues WEDGE Tax wedge on gross wages YPOT Potential output Exogenous Variables not Controllable by Slovenian Policy Makers BANKCAP Capital injections into the banking sector, mill. euro D1997 Dummy, 1 in 1997, 0 else D1998 Dummy, 1 in 1998, 0 else D1999 Dummy, 1 in 1999, 0 else D2000 Dummy, 1 in 2000, 0 else D2001 Dummy, 1 in 2001, 0 else 288 ECONOMIC AND BUSINESS REVIEW | VOL. 20 | No. 2 | 2018 D2002 Dummy, 1 in 2002, 0 else D2003 Dummy, 1 in 2003, 0 else D2004 Dummy, 1 in 2004, 0 else D2005 Dummy, 1 in 2005, 0 else D2008 Dummy, 1 in 2008, 0 else D2009 Dummy, 1 in 2009, 0 else D2010 Dummy, 1 in 2010, 0 else D2012 Dummy, 1 in 2012, 0 else D2013 Dummy, 1 in 2013, 0 else D2014 Dummy, 1 in 2014, 0 else D199xQi Dummy, 1 in quarter i of year 199x, 0 else D200xQi Dummy, 1 in quarter i of year 200x, 0 else DEBTADJ Change in debt level, not due to budget balance or bank capitalisation DEPR Capital stock depreciation rate EUR10Y 10 year government bond yield, Euro Area average EUR3M 3-month EURIBOR EURUSD Exchange rate, US dollar per euro EXPREST Remaining government expenditures GN_REST Public consumption, diff. between national account and fiscal stat. INVENTR Real changes in inventories OIL Oil price, USD per barrel Brent NAIRU_EU Non-accelerating inflation rate of unemployment, published by the EU Commission POP1564 Population, 15 to 64 years POP65PLUS Population, 65 years or older q1 Dummy, 1 in the first quarter of each year, 0 else REVREST Remaining government revenues SITEUR Exchange rate, euro per Slovenian tolar TAXDIRRATE Other direct taxes in relation to nominal GDP TAXINDIRRATE Other indirect taxes in relation to nominal GDP WTRADE World trade, CPB Policy Instruments GERD Public expenditures, Research & Development GINVN Public investment, nominal GNFIN Public consumption according to fiscal statistics, nominal INCTAXRATE Average personal income tax rate LFTERSHARE Active working population with tertiary education, % of total SOCEMPRATE Average social security contribution rate TRANSFERSN Transfers to individuals and households VATAXRATE Value added tax rate 289 K. WEYERSTRASS, R. NECK, D. BLUESCHKE, B. MAJCEN, A. SRAKAR, M. VERBIČ | SLOPOL10 ... The following table shows the detailed results of the stationarity tests. W e report the results of Augmented Dickey-Fuller tests (ADF), Phillips-Perron tests (PP) and Kwiatkowski- Phillips-Schmidt-Shin tests (KPSS) for stationarity. The decision on lag length was based on the Schwarz information criterion (SIC). The bandwidth was automatically selected using the Newey-West (1994) approach. We used the test model with a constant and without a deterministic trend. *, **, *** denote rejection of the null hypothesis of a unit root at the 10, 5, 1 percent level of significance respectively. +, ++, +++ denote rejection of the null hypothesis of no unit root at the 10, 5, 1 percent level of significance respectively. Table A2: Results of Tests for Stationarity Levels Variable ADF Lags PP Bandwidth KPSS Bandwidth agwn -1.773 4 -1.406 13 1.127+++ 7 agwr -3.043** 4 -5.638*** 2 0.174 6 balance -1.499 3 -5.872*** 2 0.789+++ 6 balancegdp -1.734 3 -6.893*** 3 0.782+++ 5 cagdp 0.899 3 -2.588* 7 0.949+++ 6 can 2.07 3 -2.632* 23 0.873+++ 6 capr -1.547 5 -1.463 6 1.115+++ 7 cdef -1.358 4 -1.237 15 1.134+++ 7 cn -1.173 4 -1.598 14 1.121+++ 7 cpi -2.596* 5 -3.661*** 8 1.218+++ 6 cr -1.747 8 -2.995* 19 1.199+++ 6 debt 3.494 0 3.778 1 0.971+++ 7 debtgdp 2.321 0 2.086 3 0.927+++ 6 demand -1.437 5 -1.404 16 1.079+++ 7 emp -1.656 4 -2.915* 16 0.348+ 6 emp1564 -2.134 4 -2.111 21 0.367+ 6 emp65plus -3.523*** 0 -3.573*** 1 0.418+ 5 expdef -0.651 4 -0.887 6 1.115+++ 7 exr -0.446 5 -0.134 14 1.128+++ 7 gap -5.023*** 4 -8.500*** 2 0.134 3 gdef -1.808 4 -1.259 14 1.127+++ 7 gdpdef -1.286 4 -1.36 16 1.138+++ 7 gdpn -1.146 6 -1.281 14 1.113+++ 7 gdpr -1.645 6 -1.762 16 1.041+++ 7 gerdr -1.581 3 -8.808*** 20 0.474++ 10 ginvr 0.121 3 -7.910*** 2 1.882+++ 0 gn -1.183 8 -1.097 14 1.112+++ 7 gov10y -1.384 1 -3.932*** 3 1.014+++ 6 gov10yr -4.225*** 1 -3.109** 2 0.224 5 gr -1.970 4 -1.625 14 1.063+++ 7 grgdpr -3.556*** 2 -2.789* 4 0.428+ 6 grypot -2.189 0 -2.172 2 0.846+++ 6 impdef -0.7 0 -0.78 3 1.051+++ 7 290 ECONOMIC AND BUSINESS REVIEW | VOL. 20 | No. 2 | 2018 Variable ADF Lags PP Bandwidth KPSS Bandwidth impr -1.314 4 -1.006 59 1.072+++ 7 income -1.318 5 -1.3 14 1.127+++ 7 incomer -2.268 5 -4.746*** 5 0.231 6 inctax -1.636 3 -4.629*** 22 1.04+++ 6 inctaxcorp -1.52 3 -4.783*** 2 0.616++ 6 inctaxpers -2.021 3 -5.053*** 29 1.196+++ 6 infl -0.944 4 -1.205 3 1.032+++ 6 interest 0.21 11 -7.885*** 1 1.338+++ 4 invdef 0.35 2 -0.343 21 1.125+++ 7 invn -2.369 4 -2.098 82 0.74+++ 6 invr -2.381 4 -2.181 82 0.433+ 6 lf -1.427 4 -2.934** 17 0.716++ 6 lf1564 -1.396 2 -1.903 26 0.752+++ 6 lf65plus -3.523*** 0 -3.573*** 1 0.418+ 5 netwagen -1.533 5 -1.479 14 1.113+++ 7 netwager -2.988** 4 -3.233** 49 0.458+ 6 oileur -1.505 0 -1.505 0 0.977+++ 7 primbalance -1.912 3 -5.552*** 3 0.549++ 6 primbalancegdp -2.03 3 -6.633*** 3 0.557++ 5 prinvr -2.124 4 -2.041 60 0.332 6 prod -2.189 7 -2.083 16 1.241+++ 6 reer -1.949 0 -2.121 1 0.741+++ 6 sitbor3m -2.687* 1 -2.103 4 0.86+++ 6 soccomp -0.961 4 -1.017 15 1.107+++ 7 socemp -1.721 4 -1.415 14 1.119+++ 7 soctotal -1.378 4 -1.221 14 1.116+++ 7 taxdirrest -2.534 4 -2.988** 20 0.629++ 6 taxindirrest -1.138 3 -1.752 26 1.134+++ 7 tgen -1.692 5 -1.343 14 1.125+++ 7 tgrn -1.822 4 -1.786 15 1.114+++ 7 trendemp -1.568 4 -3.151** 13 0.575++ 6 trendtfp -1.877 8 -5.521*** 6 1.009+++ 7 ucc -4.266*** 1 -3.154** 2 0.216 5 ulc -1.500 4 -1.549 19 1.033+++ 7 un -2.472 8 -1.639 5 0.483++ 7 un1564 -2.306 8 -1.505 5 0.553++ 6 ur -2.406 8 -1.717 7 0.408+ 7 ur1564 -2.472 8 -1.611 6 0.464++ 6 util -5.023*** 4 -8.500*** 2 0.134 3 vat -1.399 3 -4.813*** 12 1.251+++ 6 wedge -2.666* 3 -2.025 16 1.127+++ 7 ypot -2.068 4 -2.094 14 1.085+++ 7 debtadj -13.689*** 0 -13.711*** 3 0.147 0 depr -0.415 4 -0.319 85 0.449+ 6 eur10y -2.193 1 -2.336 4 1.067+++ 6 291 K. WEYERSTRASS, R. NECK, D. BLUESCHKE, B. MAJCEN, A. SRAKAR, M. VERBIČ | SLOPOL10 ... Variable ADF Lags PP Bandwidth KPSS Bandwidth eur3m -2.414 1 -1.855 4 0.988+++ 6 eurusd -2.035 1 -1.624 2 0.382+ 6 exprest -0.89 4 -2.477 19 1.147+++ 7 gerd -1.504 3 -8.284*** 7 1.362+++ 0 ginvn 0.469 3 -7.201*** 0 1.552+++ 3 gn_rest -0.316 3 -4.877*** 4 0.565++ 6 gnfin -2.125 4 -1.784 15 1.09+++ 7 inctaxrate -3.075** 3 -7.214*** 1 0.942+++ 5 inventr -3.137** 4 -5.843*** 1 0.228 5 lftershare 2.803 4 3.037 4 1.123+++ 6 nairu_eu -0.733 9 -0.807 4 1.164+++ 7 oil -1.557 2 -1.616 3 0.863+++ 7 pop1564 -0.521 5 -0.133 4 0.287 6 pop65plus 0.112 1 2.799 30 1.189+++ 6 revrest -0.709 3 -4.133*** 13 1.336+++ 6 siteur -2.689* 8 -7.179*** 9 0.901+++ 7 socemprate -3.082** 4 -5.357*** 42 1.108+++ 6 taxdirrate -1.929 4 -2.733** 36 0.249 6 taxindirrate -1.487 3 -3.223** 8 0.954+++ 6 transfersn -2.19 4 -1.663 14 1.175+++ 7 vataxrate -1.729 3 -11.539*** 2 0.656+++ 27 wtrade -1.029 2 -0.938 1 1.185+++ 7 ypot -2.068 4 -2.094 14 1.085+++ 7 First Differences Variable ADF Lags PP Bandwidth KPSS Bandwidth agwn -2.312 3 -33.323*** 47 0.254 13 agwr -2.334 3 -31.946*** 28 0.096 13 balance -13.39*** 2 -28.624*** 17 0.109 15 balancegdp -14.273*** 2 -30.893*** 16 0.104 15 cagdp -11.625*** 2 -22.159*** 19 0.303 18 can -5.417*** 3 -15.823*** 17 0.338 16 capr -1.864 4 -2.287 51 0.398+ 6 cdef -3.172** 3 -11.877*** 14 0.192 14 cn -2.898** 3 -21.676*** 13 0.142 13 cpi -0.838 3 -8.512*** 2 1.28+++ 2 cr -2.123 7 -28.605*** 14 0.218 13 debt -4.499*** 1 -8.642*** 4 0.709++ 5 debtgdp -4.478*** 1 -8.394*** 4 0.495++ 5 demand -3.641*** 4 -21.409*** 42 0.185 15 emp -3.816*** 3 -10.045*** 26 0.128 25 emp1564 -3.727*** 3 -9.087*** 27 0.165 29 emp65plus -9.544*** 0 -12.997*** 14 0.157 17 292 ECONOMIC AND BUSINESS REVIEW | VOL. 20 | No. 2 | 2018 Variable ADF Lags PP Bandwidth KPSS Bandwidth expdef -3.273** 3 -9.309*** 7 0.072 7 exr -4.754*** 4 -9.687*** 12 0.098 15 gap -5.356*** 6 -42.042*** 23 0.128 13 gdef -2.872* 3 -21.594*** 27 0.176 14 gdpdef -3.353** 3 -13.965*** 17 0.221 15 gdpn -3.437** 5 -17.76*** 16 0.148 13 gdpr -4.001*** 5 -19.49*** 33 0.216 14 gerdr -28.757*** 2 -20.675*** 13 0.091 12 ginvr -40.618*** 2 -24.808*** 13 0.16 13 gn -1.841 7 -27.178*** 4 0.151 13 gov10y -2.888* 10 -12.684*** 3 0.333 8 gov10yr -7.119*** 0 -7.091*** 3 0.089 3 gr -2.279 3 -29.073*** 2 0.195 14 grgdpr -5.946*** 3 -8.009*** 3 0.037 3 grypot -9.439*** 0 -9.449*** 2 0.037 2 impdef -8.791*** 0 -8.840*** 3 0.084 3 impr -3.214** 3 -13.062*** 10 0.23 37 income -2.802* 4 -14.353*** 14 0.14 13 incomer -2.717** 4 -14.622*** 14 0.079 14 inctax -12.354*** 2 -31.134*** 19 0.165 13 inctaxcorp -13.754*** 2 -25.119*** 16 0.113 14 inctaxpers -15.093*** 2 -44.113*** 17 0.175 13 infl -6.092*** 3 -6.855*** 3 0.036 3 interest -3.058** 10 -29.74*** 13 0.101 13 invdef -12.284*** 1 -9.487*** 27 0.11 20 invn -2.602* 3 -12.377*** 18 0.246 23 invr -2.753* 3 -13.303*** 46 0.272 19 lf -11.16*** 1 -10.608*** 26 0.15 25 lf1564 -10.165*** 1 -10.062*** 27 0.164 29 lf65plus -9.544*** 0 -12.997*** 14 0.157 17 netwagen -2.883* 4 -20.567*** 14 0.156 13 netwager -3.306** 3 -16.111*** 14 0.124 13 oileur -7.438*** 0 -7.351*** 3 0.179 0 primbalance -10.064*** 2 -37.165*** 40 0.149 20 primbalancegdp -11.229*** 2 -35.294*** 25 0.131 18 prinvr -2.938** 3 -10.627*** 19 0.358+ 18 prod -5.074*** 6 -24.469*** 25 0.287 14 reer -7.864*** 0 -7.904*** 1 0.047 1 sitbor3m -6.426*** 0 -6.414*** 1 0.083 4 soccomp -4.44*** 3 -22.854*** 26 0.124 14 socemp -2.726 4 -23.800*** 23 0.199 13 soctotal -3.8 3 -23.724*** 23 0.169 13 taxdirrest -3.387 3 -14.619*** 15 0.328 14 293 K. WEYERSTRASS, R. NECK, D. BLUESCHKE, B. MAJCEN, A. SRAKAR, M. VERBIČ | SLOPOL10 ... Variable ADF Lags PP Bandwidth KPSS Bandwidth taxindirrest -15.542 2 -29.294*** 17 0.19 15 tgen -2.794 4 -33.417*** 14 0.116 13 tgrn -5.585 3 -41.022*** 15 0.166 13 trendemp -11.161 1 -10.692*** 26 0.15 25 trendtfp -1.712*** 7 -1.668 6 0.767+++ 7 ucc -7.164*** 0 -7.137*** 3 0.085 3 ulc -2.849* 3 -17.118*** 32 0.163 15 un -1.853 7 -9.096*** 9 0.082 10 un1564 -2.713* 3 -8.385*** 8 0.11 9 ur -2.029 7 -9.325*** 12 0.086 14 ur1564 -1.572 7 -8.359*** 11 0.112 13 util -5.356*** 6 -42.042*** 23 0.128 13 vat -19.866*** 2 -42.366*** 14 0.094 13 wedge -5.984*** 3 -42.232*** 15 0.197 13 ypot -2.609* 3 -8.314*** 8 0.555++ 6 debtadj -8.254 5 -36.099*** 5 0.114 17 depr -9.447 3 -9.466*** 26 0.361+ 19 eur10y -6.358 0 -6.291*** 2 0.207 4 eur3m -5.024 0 -5.099*** 1 0.063 4 eurusd -6.762 1 -6.323*** 8 0.131 3 exprest -6.328 3 -25.289*** 13 0.084 13 gerd -28.241 2 -21.678*** 13 0.063 13 ginvn -44.566 2 -27.355*** 13 0.175 13 gn_rest -22.335 2 -24.487*** 14 0.237 13 gnfin -2.573 3 -29.785*** 55 0.213 13 inctaxrate -22.203 2 -37.677*** 14 0.187 13 inventr -4.443 3 -24.159*** 22 0.108 15 lftershare -2.365 3 -7.962*** 1 0.909+++ 3 nairu_eu -3.005 8 -4.262*** 2 0.062 4 oil -7.291 1 -6.852*** 9 0.159 4 pop1564 -2.873 4 -8.365*** 4 0.508++ 4 pop65plus -13.868 0 -14.307*** 8 0.489++ 47 revrest -17.644 2 -38.455*** 14 0.082 14 siteur -2.372 7 -6.142*** 4 1.02+++ 5 socemprate -3.622 3 -25.702*** 13 0.252 13 taxdirrate -2.925 3 -10.84*** 28 0.277 18 taxindirrate -14.309 2 -27.146*** 20 0.131 15 transfersn -3.346 4 -26.334*** 17 0.346 13 vataxrate -19.501 2 -50.457*** 14 0.098 13 wtrade -5.956 1 -4.453*** 9 0.061 1 294 ECONOMIC AND BUSINESS REVIEW | VOL. 20 | No. 2 | 2018 The following table shows the results of the cointegration tests for the behavioural equations finally adopted. *, **, *** means that the null hypothesis (ADF and Phillips-Perron: no stationarity of the residuals; KPSS: stationarity of the residuals) can be rejected at the 10, 5, 1 percent level of significance respectively. Similarly to the tests for stationarity, we chose the models with a constant, but without a trend. As before, the decision on lag length was based on the Schwarz information criterion. The bandwidth was selected automatically using the Newey-West (1994) approach. Table A3: Tests for Cointegration – Tests for Stationarity of Residuals of the Equations Equation ADF Lags PP Bandwidth KPSS Bandwidth Trend TFP -2.012 4 -3.872*** 5 0.176 6 Consumption -6.536*** 0 -6.546*** 3 0.065 2 Investment -7.636*** 0 -7.913*** 5 0.195 5 Exports -7.243*** 0 -7.267*** 1 0.092 1 Imports -9.165*** 0 -9.156*** 4 0.124 4 Employment 15-64 -4.250*** 0 -4.250*** 0 0.184 4 Employment 65+ -7.983*** 0 -7.984*** 1 0.109 2 Labour supply 15-64 -5.241*** 0 -5.260*** 1 0.264 3 Labour supply 65+ -7.965*** 0 -7.965*** 1 0.098 1 Wage rate -8.002*** 0 -7.999*** 1 0.060 0 CPI -6.739*** 0 -6.806*** 2 0.048 3 Cons. Deflator -5.007*** 0 -5.039*** 2 0.082 3 Gov. cons. deflator -8.062*** 0 -8.062*** 0 0.093 1 Investment deflator -4.739*** 0 -4.739*** 0 0.217 4 Export deflator -6.105*** 1 -6.288*** 4 0.074 2 Import deflator -5.127*** 3 -4.563*** 5 0.124 5 Short-term int. rate -5.080*** 0 -5.080*** 0 0.086 4 Long-term int. rate -3.865*** 5 -4.357*** 4 0.205 4 Real eff. exch. rate -4.592*** 0 -4.550*** 2 0.131 5 Soc. sec. revenues -7.798*** 0 -7.869*** 3 0.130 4 Company taxes -9.062*** 0 -9.161*** 5 0.105 5 V AT revenues -2.920** 3 -8.474*** 8 0.175 3 Interest payments -9.239*** 0 -9.244*** 2 0.216 2 295 K. WEYERSTRASS, R. NECK, D. BLUESCHKE, B. MAJCEN, A. SRAKAR, M. VERBIČ | SLOPOL10 ... Table A4: Results of Ex-post Model Evaluation Variables in levels Variable RMSPE Theil MAPE Variable RMSPE Theil MAPE AGWN 4.1 0.359 3.6 INTEREST 9,463.4 0.660 18.1 AGWR 2.0 0.516 1.8 INVDEF 1.8 0.459 1.2 BALANCE 247.8 0.689 293.7 INVN 10.6 0.814 8.6 CAN 467.9 1.062 447.9 INVR 11.0 0.838 9.2 CAPR 7.2 0.373 6.5 LF 0.9 0.767 0.7 CDEF 2.0 0.570 1.5 LF1564 0.9 0.795 0.6 CN 5.1 0.543 4.2 LF65PLUS 9.4 0.726 7.2 CPI 4.4 0.436 3.3 NETW AGEN 4.1 0.369 3.6 CR 3.2 0.557 2.7 NETW AGER 2.0 0.381 1.8 DEBT 22.8 0.160 21.1 OILEUR 0.0 0.000 0.0 DEMAND 2.0 0.328 1.6 PRIMBALANCE 9,081.8 0.679 339.0 EMP 1.4 0.787 1.3 PRINVR 12.3 0.854 10.4 EMP1564 1.3 0.778 1.2 PROD 2.0 0.610 1.7 EMP65PLUS 16.2 1.034 12.2 REER 2.2 0.697 1.9 EXPDEF 0.8 0.484 0.7 SOCCOMP 5.2 0.430 4.6 EXR 2.1 0.197 1.7 SOCEMP 4.5 0.387 3.9 GDEF 2.0 0.431 1.7 SOCTOTAL 4.8 0.392 4.2 GDPDEF 8.2 0.366 0.8 TAXDIRREST 2.9 0.257 2.5 GDPN 2.8 0.513 2.4 TAXINDIRREST 3.0 0.366 2.6 GDPR 2.3 0.525 1.9 TGEN 0.5 0.056 0.4 GERDR 1.6 0.054 1.2 TGRN 3.8 0.458 3.0 GINVR 1.8 0.080 1.4 TRENDEMP 0.9 0.759 0.7 GN 0.0 0.000 0.0 TRENDTFP 3.8 1.164 0.0 GR 1.9 0.532 1.6 UCC 49.4 1.134 40.9 IMPDEF 1.7 0.451 1.5 ULC 3.6 0.682 3.0 IMPR 4.4 0.418 3.8 UN 18.7 1.044 15.9 INCOME 2.5 0.463 2.1 UN1564 17.1 0.896 14.9 INCOMER 5.2 0.621 3.8 VAT 7.2 0.653 5.7 INCTAX 8.8 0.699 7.4 WEDGE 4.1 0.250 3.6 INCTAXCORP 32.4 0.955 27.0 YPOT 5.8 0.639 5.5 INCTAXPERS 4.6 0.296 4.0 296 ECONOMIC AND BUSINESS REVIEW | VOL. 20 | No. 2 | 2018 Variables in percent Variable RMSE Theil MAE BALANCEGDP 1.4 0.777 1.0 CAGDP 1.7 1.121 1.4 DEBTGDP 7.8 0.324 7.3 GAP 5.7 0.971 4.9 GOV10Y 0.6 0.471 0.5 GOV10YR 1.8 1.140 1.5 GRGDPR 2.1 0.695 1.6 GRYPOT 1.9 1.706 1.5 INFL 1.9 0.862 1.6 PRIMBALANCEGDP 1.5 0.758 1.2 SITBOR3M 1.0 0.828 0.7 UR 1.3 1.030 1.1 UR1564 1.2 0.892 1.0 UTIL 5.7 0.969 4.9 Figure A1: Real GDP 26 Figure A2: Potential GDP Figure A3: Real GDP Growth 26 Figure A2: Potential GDP Figure A3: Real GDP Growth 27 Figure A4: Real private consumption Figure A5: Real investment 297 K. WEYERSTRASS, R. NECK, D. BLUESCHKE, B. MAJCEN, A. SRAKAR, M. VERBIČ | SLOPOL10 ... Figure A2: Potential GDP Figure A3: Real GDP Growth 26 Figure A2: Potential GDP Figure A3: Real GDP Growth 26 Figure A2: Potential GDP Figure A3: Real GDP Growth 27 Figure A4: Real private consumption Figure A5: Real investment 298 ECONOMIC AND BUSINESS REVIEW | VOL. 20 | No. 2 | 2018 Figure A4: Real private consumption Figure A5: Real investment 28 Figure A6: Consumer Price Index Figure A7: Inflation Rate 28 Figure A6: Consumer Price Index Figure A7: Inflation Rate 29 Figure A8: Employment Figure A9: Unemployment Rate 27 Figure A4: Real private consumption Figure A5: Real investment 299 K. WEYERSTRASS, R. NECK, D. BLUESCHKE, B. MAJCEN, A. SRAKAR, M. VERBIČ | SLOPOL10 ... Figure A6: Consumer Price Index Figure A7: Inflation Rate 28 Figure A6: Consumer Price Index Figure A7: Inflation Rate 29 Figure A8: Employment Figure A9: Unemployment Rate 300 ECONOMIC AND BUSINESS REVIEW | VOL. 20 | No. 2 | 2018 Figure A8: Employment Figure A9: Unemployment Rate 31 Figure A12: Net Exports in relation to Nominal GDP 30 Figure A10: Public Debt in relation to Nominal GDP Figure A11: Budget balance in relation to Nominal GDP 30 Figure A10: Public Debt in relation to Nominal GDP Figure A11: Budget balance in relation to Nominal GDP 29 Figure A8: Employment Figure A9: Unemployment Rate 301 K. WEYERSTRASS, R. NECK, D. BLUESCHKE, B. MAJCEN, A. SRAKAR, M. VERBIČ | SLOPOL10 ... Figure A10: Public Debt in relation to Nominal GDP Figure A11: Budget balance in relation to Nominal GDP 31 Figure A12: Net Exports in relation to Nominal GDP 30 Figure A10: Public Debt in relation to Nominal GDP Figure A11: Budget balance in relation to Nominal GDP 29 Figure A8: Employment Figure A9: Unemployment Rate 302 ECONOMIC AND BUSINESS REVIEW | VOL. 20 | No. 2 | 2018 Figure A12: Net Exports in relation to Nominal GDP 31 Figure A12: Net Exports in relation to Nominal GDP