In this dissertation, I present three essays on international business cycles. In the first essay, I document the empirical regularities of international business cycles using the OECD Quarterly Data, and review the existing literatures in this field. By checking the data, I point out 1) net exports-output ratios both in nominal and real terms are countercyclical before 1990 for most of the OECD countries, 2) but the ratios changes their signs from negative to positive after 1990 for some of the countries, and 3) the main reason for the sign changes is that there are changes in the relationship between exports and output: exports were weakly correlated with output or were lagged with output before 1990, but exports become strongly correlated with output and also coincident. In the literature review part, I suggest that many of the properties of international real business cycles can be accounted for by benchmark international real business cycle models, such as Backus, Kehoe and Kydland (1992) and subsequent literatures, but those models cannot account for the coexistence of procyclical and countercyclical net exports. Further, incorporating Bansal and Yaron (2004)-style multi-factor productivity with short-run (trend-stationary transitory) shocks and long-run (difference-stationary growth) shocks are promising in order to account for the new observation about the trade variables. In the second essay, I document that the correlation between net exports and output has not always been negative after 1960. For the G6 countries, most of the countries experienced countercyclical net exports before 1990. However, some of these countries, including Germany and Japan, experienced procyclical net exports after 1990 even though they experienced countercyclical net exports before that. I also show that a simple one-good two-country business cycle model with a multi-factor productivity process can explain the phenomena. A positive transitory shocks to productivity leads to a positive response in net exports because its consumption risk-sharing effect, which causes a international resource flow from Home to Foreign country, is larger than its efficiency effect, which causes an increase in investments in Home country by importing goods form Foreign country. On the other hand, a positive growth shocks to productivity lead to a negative response in net exports because its consumption risk-sharing effect is smaller than its efficiency effect. I estimate the stochastic productivity processes for the G6 countries by using the simulated method of moments, and the simulation results of the model based on the estimated parameters are able to account for the changes in net export dynamics from pre-1990 to post-1990 for Germany and Japan. In the third essay, I document that there are changes in the correlations about trade variables and capital flows for the G7 countries: 1) the magnitude of the contemporaneous correlation of exports with output is a half of that of imports with output for pre-1990, but the former is almost the same value as the latter for post-1990, 2) the magnitude of the contemporaneous correlation of real net exports-output ratio with output is significantly negative for pre-1990, but it becomes almost zero or weakly positive for post-1990. I present two types of two-country two-good real business cycle models, one of which is with complete financial markets and the other one is with incomplete financial markets model in a sense that only risk-free one-period bonds are traded. I also add two types of shocks, transitory and growth shocks, to these two models in the spirit of Aguiar and Gopinath (2007). Firstly, the standard complete financial markets model has a strong correlation of exports with output and a weak correlation of imports with output. Secondly, the standard incomplete financial markets model has a weak correlation of exports with output and a strong correlation of imports with output. Finally, with reasonable changes in model parameter values, both the complete and incomplete market models can account for the two empirical regularities above, but only the incomplete market model can account for the empirical regularities for pre-1990. I evaluate these models in light of cross-country correlation properties based on actual data, especially the cross-country consumption correlation anomaly. I show that the incomplete financial markets model is still better than the complete market model because the cross-country consumption correlation in the incomplete financial markets model is still larger than but closer to the cross-country output correlation compared with the case of the complete financial markets model.