Background: This paper presents a machine learning approach to forecast Croatia’s international bilateral trade. Objectives: The goal of this paper is to evaluate the performance of machine learning algorithms in predicting international bilateral trade flows related to imports and exports in the case of Croatia. Methods/Approach: The dataset on Croatian bilateral trade with over 180 countries worldwide from 2001 to 2019 is assembled using main variables from the gravity trade model. To forecast values of Croatian bilateral exports and imports for a horizon of one year (the year 2020), machine learning algorithms (Gaussian processes, Linear regression, and Multilayer perceptron) have been used. Each forecasting algorithm is evaluated by calculating mean absolute percentage errors (MAPE). Results: It was found that machine learning algorithms have a very good predicting ability in forecasting Croatian bilateral trade, with neural network Multilayer perceptron having the best performance among the other machine learning algorithms. Conclusions Main findings from this paper can be important for economic policymakers and other subjects in this field of research. Timely information about the changes in trends and projections of future trade flows can significantly affect decision-making related to international bilateral trade flows.

This paper provides a detailed empirical study of trade creation and trade diversion effects arising from Croatia's two regional trade agreements, the Central European Free Trade Agreement (CEFTA) and the European Union (the EU). It offers a foundation for discussion about future trade policies in terms of benefits and drawbacks from those regional trade agreements. Croatia’s imports, exports and total trade flows with 180 trading partner countries were examined for the period of 2000 – 2016. Cross-country panel regression using gravity model of international trade assessed pooled OLS, fixed and random effects, as well as more robust Tobit and PPML estimator models. The random effects model found positive effects of Croatia-CEFTA integration evident in trade creation in imports, exports and total trade flows. Croatia-EU integration exhibits no significant effect of trade creation in neither imports, exports nor total trade flows. Nonetheless, there is a trade diversion effect in cases of imports and total trade flows. In the Tobit model CEFTA created trade in imports, exports and total trade flows, while the EU diverted trade in imports and total trade flows. Finally, the robust PPML estimator found that: (1) CEFTA membership created trade in imports, exports and total trade flows, and (2) the EU membership diverted trade in imports and exports, and created trade in total trade flows.

This paper investigates trends and determinants in intra-industry trade in Croatia. Intra-industry trade refers to a two-way trade of differentiated products. The level of intra-industry trade between Croatia and EU, CEE, and countries worldwide is calculated using the Grubel-Lloyd index under HS-2 and HS-4 product groups. Dynamic changes in intra-industry trade are shown by calculating marginal intra-industry trade indices. In addition to this, bilateral weighted adjusted intra-industry trade indices for Croatia and 24 most important trading partner countries are calculated in the period from 2001 to 2017. A panel-econometric analysis investigates the determinants of intra-industry trade in Croatia using a random effects model due to time-invariant characteristics of some determinants. The results of the analysis show that gross domestic product, common border, EU membership, and the distance between Croatia and trading partner countries have significant influence on the level of intra-industry trade in Croatia.

Economists have argued about the relationship between trade openness and economic growth for a very long time. Despite garnering a lot of attention in the literature for more than a century, we still don't have a satisfying response to this subject. This study aims to examine the relationship between trade openness and GDP growth in the case of OECD countries, in the period from 1988 to 2020. In order to tackle data heterogeneity across countries and to form a robust (outlier resistant) regression, a quantile regression is used. Furthermore, panel dynamic GMM was used due to “more countries than time-period case“. Main findings of the paper validate the hypothesis that trade openness facilitates economic growth. Most of the explanatory variables have been significant in the regression models obtaining predicted signs of the regression coefficients, in accordance with the economic theory. The quantile regression modelling showed discrepancies between different quantiles. For tau equal to 0.25 the value of regression coefficient related to trade openness variable was negative.

Zipf’s law is a striking regularity in the field of urban economics that states that the sizes of cities should follow the rank-size distribution. Rank-size distribution, or the rank-size rule, is a commonly observed statistical relationship between the population size and population rank of a nations’ cities. The goal of this paper is to test Zipf’s law as applied to data for settlements and cities in Croatia using the Census of Population Survey for the year 2011. The results of the analysis have shown that Zipf’s law for settlements in Croatia holds true for the majority of the settlement sizes. However, the rank-size distribution does not hold true for extremely small and extremely large settlement sizes. When city proper and urban agglomeration of 127 Croatian cities were examined, Zipf’s law was found to hold true only for urban agglomerations. The results of the study are discussed in terms of regional development.

This paper presents empirical evidence on the validity of the Linder hypothesis in the case of Croatia. According to the Linder hypothesis, one of the new theories of international trade, countries with a similar level of income per capita should trade more. In order to investigate the trade pattern of Croatia's international trade, a panel regression model is formulated including 184 Croatia's import partner countries in the period from 2000 to 2016. The Linder effect was displayed and calculated using the Linder variable expressed as an absolute difference between GDP per capita of the importing and the exporting country. The cross-country panel regression model is estimated using Pooled OLS, Fixed and Random effects models. Results of the analysis have shown that the validity of the Linder hypothesis for Croatia cannot be accepted. Instead, the structure of Croatia's trade is in line with the gravity model of international trade.