3 SUMMER 2024 NO.86 / THOUGHT LEADERSHIP BRIEF Our key dependent or outcome variable is yield spread, expressed as the difference between the coupon rate of one ABS tranche and a benchmark rate, which is the Chinese government bond yield with a similar duration. The smaller the difference, the better the ABS pricing. Since the differences vary drastically across ABS deals, we therefore take a natural log of such percentage differences to minimise the skewness of distribution. Blockchain We construct a dummy variable which takes value 1 when an ABS is issued based on blockchain technology, and zero otherwise. We gather qualitative evidence from various sources to determine whether a focal ABS is blockchain-based or not. We first comb the data downloaded from WIND to identify names of the originators and underwriters. We then search key words such as blockchain and ABS among corporate news or annual reports of public firms or banks. We also triangulate our search results with other sources of information such as analyst reports, teaching cases and white papers issued by major technology firms. Prior shared experience (PSE) We measure PSE using a relational dyadic network approach. In particular, we consider all possible dyads consisting of two parties among N parties in an ABS deal (eg, issuer, auditor, rating agency, etc), count the number of prior joint transactions for each dyad and calculate the average number of transactions across all dyads, which is the PSE score. Figure 2 provides a visual representation of how we calculate the PSE score. To address potential endogeneity, we adopt a coarsened exact matching (CEM) approach to filter our sample observations. CEM is a matching method to deal with endogeneity commonly observed in archive-based studies. CEM essentially matches each treatment observation (i.e. a blockchain ABS deal) with one or several comparable control observations (i.e. a non-blockchain ABS deal) on several key visible dimensions such as the total amount of principal. We also separately consider trading in the primary market and that in the secondary market. Like IPO in stock markets, an ABS deal is often traded more actively at issuance or in primary trading. IMPLICATIONS Our study finds that overall blockchain adoption significantly improves ABS pricing as the conventional wisdom predicts, as adopting blockchain is indeed associated with a better yield spread for ABS products. Specifically, the yield spread by approximately 25 basis points and that this benefit is heterogeneous across the different underlying asset classes and institutional arrangements but remain relatively the same across primary and secondary trading. This effect is robust to several model specifications. Moreover, the blockchain effect is heterogeneous across different institutional arrangements and asset classes. Figure 2. An Illustrative Example of PSE Measurement Notes: Assuming that an ABS deal involves five parties (ie I, J, K, L and M), we have 10 dyads or links. In this example the dyad I-J has three prior transactions, the dyad J-K two prior transactions and the other dyads zero prior transactions. Then, the PSE score, a measure of familiarity for this group of parties, is (3+2)/10=0.5. 2 3 Party K Party J Party I Party M Party L
RkJQdWJsaXNoZXIy NDk5Njg=