Abstract
We propose to measure the systemic risk in the shadow banking sector. Instead of testing how many institutions will fail due to the initial breakdown of one institution as extant network models do, we associate the systemic risk of one shadow banking sector with the total amount of unexpected losses it might generate both directly and indirectly. Our model focuses on balance sheet contagion and applies a loop algorithm to risk transfer. The result shows that trust companies were the main culprit of financial instability and commercial banks assumed the main risks over 2007–12 in the Chinese shadow banking system.
Notes
1. Upper and Worms (Citation2004) argue that the independence matrix X has the unappealing feature that the maximization of entropy creates elements on the main diagonal that are nonzero if a bank is both lender and borrower. In that case, using X to compute bilateral exposures would amount to assuming that banks lend to themselves. Thus, they make an additional estimation to turn the elements on the main diagonal to zero. In this article, however, we don’t face the same problem because we measure systemic risk at the sector level. In the shadow banking system, a certain sector actually includes many homogeneous companies that conduct similar business. Because companies in the same sector may lend to or borrow from each other, the sector can be both the debtor and the creditor. It is not abnormal to find the elements on the main diagonal that are nonzero; hence, we do not have to turn them to zero.