The role of non-bank credit in Credit-less recoveries

Since Calvo et al. (2006) introduced the concept of “Phoenix Miracle” related to Systemic Sudden Stop in emerging markets, several studies have been conducted regarding this phenomenon named credit-less recovery [CLR]. These studies stated how recoveries in business cycle have occurred without a pick-up in real bank credit to the private sector (Sugawara et al., 2013) with findings in developing, emerging and advanced economies. Existing literature described credit-less events relying on bank credit to private sector and it tried to understand main determinants of this credit absence basing on features of both preceding expan-sion phases and recessions. In contrast, the study aims firstly at examining the ambiguity related to these events, which are defined as “credit-less” considering only bank lending so far. The BIS da-tabase provides an unexplored scenario where recoveries can be split in four categories: bank-with/credit-with [BWCW], bank-with/credit-less [BWCL] (i.e. credit-with in literature), banking-less/credit-with [BLCW] and banking-less/credit-less recovery [BLCL], which were both consid-ered as CLR in previous studies. Moreover, the frequency of each type in different country groups is computed (emerging vs developed countries). Therefore, an empirical analysis is needed to as-sess how different compositions of credit stock can affect recoveries, since for example BLCW shows a negative bank credit flow but a positive total one. According to this, the second goal means to compare these three different types of recovery in terms of GDP components. Basing on previous studies, we know that CLR are much less effective in terms of output due to financial frictions that hamper investments (Calvo et al., 2006). However, the average results attached to these episodes included both BLCL and BLCW, therefore this new study could result in a less sig-nificant difference in output and its components between BLCW and credit-with, giving a first in-sights regarding interaction between economic and credit cycle, and a more profound one between BLCL and the other types. Given this, the third goal is the assessment of BLCL. In particular, em-pirical results of the previous two points are summarized in order to verify if BLCL are subjected to clustering effect, which means they are concentrated geographically and/or around periods characterized by global shock (Abiad et al. 2011). Finally, BLCL are further explored in relation to credit flow dynamic, to see whether nonbank’s credit diverge from banking one in most of BLCL. This can give the second insight regarding the interaction between the two cycles, specifically the substitutability between banking and non-banking channels. In summary, these analyses might an-swer the question raised by Bernanke (2007), at least with reference to recovery phase. “Does the rise of nonbank lenders make the bank-lending channel irrelevant?”