50 X.Zhang et aL /Physica A 415 (2014)43-53 05 October 2008 0.3 045 October 2008 04 025 f 02 02 01 V 2004-22005-122000-122007-12200-2.2009-122010-122011-2202-12 2004-122005-122005-22007-123099-122009-122010-22011-22012-12 Tme October 2008 October 2008 2004-122005-123006-122007-122009-122009-122010-122011-122012-12 Tme 2004-122005-122006-122007-122000-12200%-122010-122011-2201212 03 d 02 October 2008 October 2008 2004-122005-122006-122007-122008-122009-122010-22011-122012-12 12 2006-12207-122009-122009-122010-22011-122012-12 October 2008 October 2008 2004-122005-122006-122007-122008-122009-122010-122011-122012-12 2004-122005-122005-122007-22008-122009-122010-422011-122012-12 Time Fig.5.Contrast of the Dynamic Granger causality index with 24-month moving window and Monthly weighted index of market price according to the capacity ratio of each ship type from December 2004 to December 2012.(a)New-building ship market,(b)second-hand ship market,(c)freight market, (d)shipping stock market. the new ship monthly market price return versus time and the trend of the new ship weighted index price made up of 12 representative time series according to the capacity market ratio of each ship type during the same time period.The DCI shows a rich dynamic behavior over the entire period from January 2003 to December 2012.During the pre-crisis period,the DCI tends to be stable and only fluctuates from about 0.1 to 0.15.Beginning in June 2008,just prior to the October 2008 crash in new ship prices,the DCI begins to exhibit a sharp increase in the number of causality links.This indicates that the entire system including the real shipping market and the shipping financial market prices has become much more interconnected, a significant systemic risk indicator,especially during the financial crisis of 2008.We further confirm our results by studying the DCl of another three markets,that are second-hand ship prices,freight market and shipping stock market,also comparing the trend of DCl to real price weighted index within the same time period.Fig.5(b)-(d)all show the largest DCI peak in October 2008 when there is a rapid drop in the real price index.The DCI shows itself to be an efficient measure of systemic risk in the shipping industry and can provide a useful early-warning signal that serious market regression will soon occur.50 X. Zhang et al. / Physica A 415 (2014) 43–53 (b) a b c d Fig. 5. Contrast of the Dynamic Granger causality index with 24-month moving window and Monthly weighted index of market price according to the capacity ratio of each ship type from December 2004 to December 2012. (a) New-building ship market, (b) second-hand ship market, (c) freight market, (d) shipping stock market. the new ship monthly market price return versus time and the trend of the new ship weighted index price made up of 12 representative time series according to the capacity market ratio of each ship type during the same time period. The DCI shows a rich dynamic behavior over the entire period from January 2003 to December 2012. During the pre-crisis period, the DCI tends to be stable and only fluctuates from about 0.1 to 0.15. Beginning in June 2008, just prior to the October 2008 crash in new ship prices, the DCI begins to exhibit a sharp increase in the number of causality links. This indicates that the entire system including the real shipping market and the shipping financial market prices has become much more interconnected, a significant systemic risk indicator, especially during the financial crisis of 2008. We further confirm our results by studying theDCI of another three markets, that are second-hand ship prices, freight market and shipping stock market, also comparing the trend of DCI to real price weighted index within the same time period. Fig. 5(b)–(d) all show the largest DCI peak in October 2008 when there is a rapid drop in the real price index. The DCI shows itself to be an efficient measure of systemic risk in the shipping industry and can provide a useful early-warning signal that serious market regression will soon occur