First you disaggregate Vlume of CO2 emissions into its drivers and sub-drivers (frist level would be "type of emission", e.g., plane emissions, ship emissions, car emissions, agriculture emissions, etc.).
Once you have set up a nice driver tree, you think from the end of each branch - how did the financial crisis influence the intensity? For example, the amount of travel certainly reduced due to higher budget restrictions in private households (leisure travel), companies (business travel), and the public sector (business travel). Trade volumes reduced, so ship emissions are probably also negatively affected. Etc.
In the end you should be able to clearly state something like: "The financial crisis has negatively impacted (decreased) global CO2 emissions via 4 levers, while it had a postive effect (increased) on CO2 emissions via 1 lever.
The quantification would then be an entire case in itself.