Discussion of Marco Arena, Carmen Reinhart, and Francisco Vazquez
The Lending Channel in Emerging Markets: Are Foreign Banks Different? Linda Goldberg Federal Reserve Bank of New York and NBER The views in this paper are those of the author and do not necessarily reflect the position of the Federal Reserve Bank of New York or the Federal Reserve System. 1
Paper overview
Empirical investigation of lending and deposits (quantity, prices) responses to macro variables
Domestic and foreign-owned banks compared Sample of banks covers Latin America and Asia 1990s overall, versus just in crisis windows
Bottom line Domestic and Foreign-Owned Banks are more similar than they are different Differences support view that foreign banks are more stable lenders
2
Main comments Carefully executed, clearly written, honestly exposited My suggestions are organized as Data related sample period limited, additional refinements needed
Conceptual Are banks behaving optimally? Have foreign-owned banks delivered benefits as expected?
Bottom line: Paper’s conclusions are consistent with findings of prior studies that focused on fewer countries with higher frequency data Additional work likely to strengthen view that foreign banks significantly stabilized and strengthened banking systems in EMs
3
Potential effects of global banking for EM countries
Benefits
Foreign bank entry leads to greater efficiency in local banking markets Improved credit and risk allocation Competition fosters development of local markets Knowledge transfer helps spread global best practice Acquired banks gain access to global capital markets
Costs
Cherry picking of clients, weakening viable local entities? Cut and run during crises? Reduced effectiveness of local monetary policy? Increased exposure to foreign shocks Ex. See discussion of BIS CGFS working group, 2004
4
Potential effects of global banking for EM countries In last 5 years, many studies of foreign versus domestic in EMs examine themes in list of benefits/ costs. Note that the list does not emphasize More lending and deposit growth in EMs More stable lending and deposits Mix of adjustment of terms versus quantities
Benchmarks for assessing “optimal” performance are needed
5
Theory needed for identifying optimality concepts and interpreting behaviors
Develop model of optimal responses of banks
Formalize financial constraints (breadth/depth of internal capital markets). Specify competitive structure and entry conditions Generate predicted optimal responses to domestic and foreign liquidity shocks
Theory provides prediction on optimal mix of response of price and quantity in loans / deposits
Observed behaviors in lending and deposits then can be viewed against concrete benchmarks. 6
Many foreign liquidity shocks likely transmitted from US and Spain.
25,000
74
Value and Number of Acquisitions of Banks in Developing Countries
68
by Source Country, 1990-2003
15,000
10,000
38 21
14
7
14
8
4
Sweden
Greece
Switzerland
7
Portugal
16
Ireland-Rep
20
Belgium
34
Canada
France
Austria
Italy
Netherlands
United Kingdom
0
Spain
32
Japan
23
Germany
22
5,000
United States
Value (in US$ Millions)
20,000
Source: Bank of England; Thompson Financial 7
Theory needed for identifying optimality concepts and interpreting behaviors
Develop model of optimal responses of banks
Formalize financial constraints (breadth/depth of internal capital markets). Specify competitive structure and entry conditions Generate predicted optimal responses to domestic and foreign liquidity shocks
Theory provides prediction on optimal mix of response of price and quantity in loans / deposits
Observed behaviors in lending and deposits then can be viewed against concrete benchmarks. 8
Comments on the data Datascope, annual series, 20 countries
Sample period reported as 1989-2001 Few observations for 1989, 1990, 1991, 2001 Really only 1992-2000. Retrieve 2001 to 2006. Longer sample provides more scope for comparing normal versus crisis response to liquidity shocks and macroeconomic cycles.
Need more filters for outlier observations, merger and acquisition years 9
Comments on the data
Goldberg et al (2000, 2001, 2002): key behavioral distinctions in Latin America are public v. private owned, and by bank health, not domestic v. foreign How do ARV count the state-owned banks?
Domestic versus Foreign taken as exogenous, but Foreign banks often enter in aftermath of crises, to help recapitalize the domestic banking system. Perhaps initially analyze foreign bank “contribution” relative to prior behavior of state-owned banks or of domestically-owned banks that were acquired?
10
EMs differ in whether foreign entry is from sale of stateowned banks (China,India) or domestic private-owned. Commercial Banks by Ownership: Regional Differences Percent share in total bank credit
100 80
15
11
11
71
78
16
6
30
31
33
18
15
34
42
13
60
20
10
2
1
11
45
1
14
48
77
21
40 5
20
5 79
55
47
17
24
73
57
56
67
80
83
34
33
13
0 1994 1999 2004 1999 2004 1994 1999 2004 1994 1999 2004 1994 1999 2004 Latin America
China and India
Private domestic banks
Other Asia
Other EMEs
Foreign-owned banks
Source: Mihaljek, BIS 2006. In Goldberg (2006)
Central Europe
State-owned banks
11
Comments on the data
When foreign banks enter, do they take over ex-ante weaker banks with more non-performing loans?
If so, foreign banks have an initial handicap. Empirics may need controls for period of balance sheet cleanup
And, sample of domestic banks in paper reflects survivorship bias (these are healthier on average than they would be absent foreign entry)
Positive contribution of foreign banks may be understated
If foreign look like surviving domestic despite these points, perhaps that is an accomplishment (again, need theory benchmarks!) 12
Overall
ARV provide a careful exploration of EM banking
If ARV expand the sample and introduce some data refinements, likely to find that foreign banks contributed more positively than reflected in current comparison with surviving domestic banks
Theory needed to provide a benchmarks against which price and quantity realizations can be judged
Studies demonstrate positive contribution of foreign banks, but could more progress have been made?
13
14
Other comments just for authors
Really need to update literature discussion and data so that paper incorporates info since 2001. I have a number of relevant papers, including in NVBER volumes and recent work with Nicola Cetorelli. We can discuss.
Quantity and bank interest rate observations are averages over portfolios, not marginal values. How will this influence interpretation of responses? (maturity structure of portfolio may be different across banks, so actual price response within year may be quite different than represented). Is there a way to construct marginal responses, rather than averages over the whole portfolio?
Should introduce shocks from parent countries of foreign banks. By internal capital markets arguments, these should matter more for affiliates.
You put a lot of energy into creating the reserve requirement database but don’t get enough mileage out of this effort. I suggest exploiting the cross-country nature of your sample, asking whether differences in policy treatment of reserve requirements and difference in foreign entry via branches or subsidiaries is associated with different ex post performance of respective banks.
Do you know if the banks in LA and Asia are all retail oriented banks? Will differences in business models for entry and activity matter for response to different shocks?
From Table 3, need filters to control for observations on loan and deposit growth below -100 percent. Also, are all the effective bank spread data reasonable? Some min/max #s strange.
Perhaps you need controls for exchange rate regimes in place? What role played by this?
BIS has a study of related papers about to be released under CGFS group.
Table 2 perhaps better reported at snapshot dates, as in Kashyup and Stein, Campello, Cetorelli and Goldberg (2007).
Perhaps, instead of separately introducing domestic liquidity and reserve requirements, construct a Bernanke-Mihov type measure for each economy. Not sure the mileage from this would be worth the effort. 15