Discussion of The International Transmission Channels of Monetary Policy Claudia Buch, Matthieu Bussiere, Linda Goldberg, and Robert Hills Jean Imbs June 2017 Imbs (2017) Banque de France - 30 June 2017 June 2017 1 / 13
IRBN Bank-level data collected for 17 European countries over 2000-2015 Up from 11 European countries over 2006-2013 (back in 2014) Balance sheet information: deposit base, capital, leverage, credit, etc. International dimension at bank level: purely domestic, foreign affi liate, parent/affi liate, net due within bank network. Wealth of information - akin to Call Reports in the US. With added cross-country dimension. Imbs (2017) Banque de France - 30 June 2017 June 2017 2 / 13
The Paper Summary of 17 country-specific studies, with analogous specifications. Main empirical question: heterogeneous effects of monetary policy shock abroad (at home) on domestic credit by international banks (foreign credit by exposed domestic banks): L b,t = α b + γ t + β 1 MP t k Channel b,t k + β 2 χ b,t 1 + ε b,t k (Btw are direct effects of MP t k and Channel b,t k controlled?) Imbs (2017) Banque de France - 30 June 2017 June 2017 3 / 13
The Channels Key variable is Channel. Many versions. Cross-border liabilities from country, Net Cross-Border liabilities, Net Intragroup Funding Cross-borders assets in country, Cross-Borders assets to banks in country, cross-border assets to non-banks in country Liquid assets, Excess reserves ratio, Securities / Total assets, Loans / total assets. International dimension key. Additional ratios more conventional. Imbs (2017) Banque de France - 30 June 2017 June 2017 4 / 13
Results Wealth of results country by country. Recurrent conclusion is β 1 = 0 and systematically significant. But the Channels that matter are not the same across countries. Idea is that monetary policy shocks are channeled within banks international network, with end effect on credit (bank channel) and on risk taking (portfolio channel). Kashyap-Stein (1994) or Cetorelli-Goldberg (2012): some banks are constrained, have limited access to alternate sources of funding because of exogenous reasons, and respond to monetary contraction with less lending. Miller-Modigliani not holding. Intuition similar to corporate finance literature on the importance of capital structure. E.g. Fazzari-Hubbard-Peterson 1988. Imbs (2017) Banque de France - 30 June 2017 June 2017 5 / 13
The Data The most disappointing aspect of the paper is on page 3: "... The tests use a common methodology applied to each country s confidential bank-level dataset". Participants analyze bank-level datasets by country and share empirical results. Not data. IRBN has the potential to spur a literature on Europe akin to what Call Reports / Home Mortgage Disclosure Act spurred on US economy. On: banking and the real economy, banking and asset prices, banking competition, predatory banking, discrimination, finance and inequality, etc. Some of these could potentially be examined at country level. But international dimension much, much richer. An enormous literature, enormously relevant, and published at highest level. Free access and panel dimension are of the essence. Imbs (2017) Banque de France - 30 June 2017 June 2017 6 / 13
Interpreting the results The key identification here is that source of heterogeneity between banks is exogenous. Kashyap-Stein: Small banks cannot raise funding easily, Miller-Modigliani does not hold, they curtail lending. Would be useful to know which "splits" the authors believe to be exogenous. There are many, many, many such splits Imbs (2017) Banque de France - 30 June 2017 June 2017 7 / 13
Heterogeneity First, there are many variables that capture Channels. Which ones are assumed to be exogenous? The most obvious exogenous Channel is the existence of an international network. Some banks simply cannot tap foreign branchs (or vice versa). But are we to believe that excess reserves, liquidity, loan portfolio,... are also exogenous? Unlikely. Imbs (2017) Banque de France - 30 June 2017 June 2017 8 / 13
Heterogeneity Argument is that individual banks decisions - even if they are endogenous - do not affect monetary policy, so that the shock itself is exogenous. What if monetary policy is synchronized internationally (2000 to 2015). What if monetary policy is concerned with credit. (2000-2015). Then, not clear that banks credit is an outcome of monetary policy. Could be an outcome of credit demand, which is a concern of monetary policy. True even if effects are heterogeneous across banks. There are many reasons why banks are heterogeneous, and they don t all imply that Miller-Modigliani fails (e.g. different risk aversion, different governance, etc). Imbs (2017) Banque de France - 30 June 2017 June 2017 9 / 13
Heterogeneity Foreign monetary policy is a residual controling for domestic monetary policy. And monetary policy shocks are also controling for lagged GDP and lagged inflation a reduced form VAR. So some of these issues are addressed - but the reader would like to know what is supposed to be exogenous more clearly. Imbs (2017) Banque de France - 30 June 2017 June 2017 10 / 13
One paper one point A lot of interesting results. Attempt to identify bank lending channel, portfolio channel, and differences in either channel between conventional and ZLB periods. I wasn t too clear whether both channels could be well identified separately (The paper mentions that it cannot be done since some Channels reflect potentially both lending and portfolio mechanisms.) I wasn t too clear either whether the differences between conventional and unconventional MP appeared in the data - and what the conclusion was. Perhaps suffi cient to focus on bank lending channel with an international dimension? Imbs (2017) Banque de France - 30 June 2017 June 2017 11 / 13
One paper one point: A suggestion For example isolate effect of foreign MP shock comparing international banks with domestic banks (presumably exogenous) Then split both groups along an exogenous variable that reflects actual constraint: size? foreign dimension? excess reserves perhaps? Then look for differential responses of endogenous variables, of which credit is only one. E.g. asset porfolio, deposits, profitability (interest income fee). Then (perhaps) look whether these differential responses vary across conventional vs. unconventional periods Imbs (2017) Banque de France - 30 June 2017 June 2017 12 / 13
Conclusion A (potentially) phenomenal dataset in Europe. Could be used to look at many things. For example, effects of financial deregulation directives on: bank profitability, assets prices, house prices, credit supply, etc. 2000-2015 especially relevant. Don t stop now. Please do everything you can to circulate data (anonymized if you must). Imbs (2017) Banque de France - 30 June 2017 June 2017 13 / 13