DSpace at VNU: Involuntary excess reserves, the reserve requirements and credit rationing in China

This article was downloaded by: [Purdue University]
On: 18 January 2015, At: 12:49
Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Applied Economics
Publication details, including instructions for authors and subscription information:
/>
Involuntary excess reserves, the reserve requirements
and credit rationing in China
a

b

Vu Hong Thai Nguyen , Agyenim Boateng & David Newton

c

a

International University, Vietnam National University, Ho Chi Minh City, Vietnam

b

Glasgow School of Business & Society, Glasgow Caledonian University, Glasgow, UK

c

Nottingham University Business School, University of Nottingham, Nottingham, UK
Published online: 07 Jan 2015.

Click for updates
To cite this article: Vu Hong Thai Nguyen, Agyenim Boateng & David Newton (2015) Involuntary excess reserves, the reserve
requirements and credit rationing in China, Applied Economics, 47:14, 1424-1437, DOI: 10.1080/00036846.2014.995362
To link to this article: />
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions


Applied Economics, 2015
Vol. 47, No. 14, 1424–1437, />
Involuntary excess reserves, the
reserve requirements and credit
rationing in China
Vu Hong Thai Nguyena, Agyenim Boatengb,* and David Newtonc

Downloaded by [Purdue University] at 12:49 18 January 2015

a

International University, Vietnam National University, Ho Chi Minh City,
Vietnam
b
Glasgow School of Business & Society, Glasgow Caledonian University,
Glasgow, UK
c
Nottingham University Business School, University of Nottingham,
Nottingham, UK

Using a sample of 95 banks that covers the period 2000–2011, this article
examines Chinese banks’ credit lending behaviour in response to the
changes in the reserve requirement ratio in the presence of involuntary
excess reserves (IERs) in the banking system. The study finds that Chinese
banks with positive IERs one period after a reserve requirement shock
experience a significantly increased credit supply in response to an
increase in reserve requirement ratio. However, the reserve requirements
have no significant impact on the credit supply in Chinese banks that have
negative IERs one period after a reserve requirement shock. This article
sheds lights on the effectiveness of Chinese monetary policy, which uses
reserve requirements as the primary tool to sterilize excess liquidity and
restrain credit expansion.
Keywords: involuntary reserve; credit rationing; Chinese banks; reserve
requirement
JEL Classification: E51; E52; E58

I. Introduction
Prior literature examining the liquidity effect of
reserve requirements on the credit supply indicates

that an increase in reserve requirement ratio drains
liquidity and reduce the credit supply (Bernanke and
Blinder, 1988; Takeda et al., 2005; Cargill and
Mayer, 2006; Mora, 2009; Gunji and Yuan, 2010).
Romer (1985) points out that increasing the reserve

requirement ratio does not only drain banking liquidity but also imposes a tax on deposits by increasing
the deposit costs for banks. Deposit cost affects credit
lending and other investment alternatives that are
available to banks such as government securities
investments (Thakor, 1996). While the funding cost
for credit lending includes both a deposit cost and a
capital requirement cost (Basel Accord), the cost of
funding for government securities investments

*Corresponding author. E-mail:

1424

© 2015 Taylor & Francis


Downloaded by [Purdue University] at 12:49 18 January 2015

Involuntary reserve, credit rationing, China
consists of only a deposit cost (Thakor, 1996). An
increase in the deposit cost reduces the capital
requirement cost’s proportion in the cost of funding
for credit lending. In other words, upon the increase
in the deposit cost, the cost of funding for credit

lending falls relative to the cost of funding for government securities investments. As a result, banks are
induced to direct investment funds from government
securities into credit lending, that is increase the
credit supply (Thakor, 1996).
The behaviour above contradicts the bank lending
channel’s argument and implies that credit supply
tends to increase in response to the increase in reserve
requirement ratio. It suggests that the liquidity effect
and the cost of funding effect of the reserve requirements operate in opposing ways, thereby making the
impact of reserve requirements on credit supply
undetermined. Despite this, the extant literature largely neglects the cost of funding effect and focuses
mainly on the liquidity effect of reserve requirement shocks (Takeda et al., 2005; Cargill and
Mayer, 2006; Mora, 2009). While a recent study by
Nguyen and Boateng (2013) examined the impact of
involuntary excess reserve (IER) on monetary policy
transmission in China, it is important to point out that
they did not analyse the impact of reserve requirements on the credit supply. Yet several studies such
as Anderson (2009), Conway et al. (2010), Ma
et al. (2011) indicate that the large excess reserves1
in the Chinese banking system is one of the reasons
behind the employment of reserve requirements by
the People’s Bank of China (PBC) as a monetary
policy tool to manage excess liquidity in China.
The main objective of this article is to examine the
Chinese banks’ credit lending behaviour in response
to the changes in the reserve requirement ratio, where
IERs which are defined as the excess reserves beyond
precautionary levels are present in the banking system (i.e. the excess liquidity situation). Examining
this behaviour is significant because credit supply is
the primary funding source in China, which drives

Chinese economic growth (Hansakul et al., 2009;
Liu and Zhang, 2010). This article therefore contributes to the bank lending literature in two important
ways. First, this study identifies the liquidity effect
and the cost of funding effect of reserve requirements
on the credit supply, which is important because
1

1425
these two effects provide conflicting predictions of
credit supply’s response. The failure to capture the
cost of funding effect may result in an unexpected
credit supply expansion after the PBC increases the
reserve requirement ratio. Second, this study sheds
lights on the impact of reserve requirements on credit
lending behaviour of Chinese banks in the context
where IERs are present. This is important because the
presence of IERs may attenuate the liquidity effect of
reserve requirements.
This study finds that Chinese banks with positive
IERs one period after a reserve requirement shock
significantly increase the credit supply in response to
an increase in reserve requirement ratio. However,
the reserve requirements have no significant impact
on the credit supply in Chinese banks that have
negative IERs one period after a reserve requirement
shock.
The remainder of the study is organized as follows:
Section II presents the theoretical background,
Section III discusses the methodology and data analysis; Section IV interprets the estimation results.
Additional analyses and robustness tests are provided

in Section V. Finally, Section VI concludes the study.

II. Theoretical Background
The credit rationing theory contends that banks
decline to screen a credit application if the net loan
benefit (the difference between loan return and credit
lending cost) fails to cover the credit screening cost
(Thakor, 1996). A higher credit lending cost reduces
the net loan benefit, which leads to an increase in the
probability of credit rationing. Thakor (1996) identifies the cost of funding and the opportunity cost
relative to alternative investments (e.g. government
securities’ return) as two components of credit lending cost (the cost to supply credit). Although the cost
of funding for credit lending includes a deposit cost
and a capital requirement cost (the cost to hold
required capital to back up bad loans), the cost of
fund for investing in government securities consists
of the only deposit cost (Thakor, 1996). An increase
in the reserve requirements raises the deposit cost
because a higher reserve requirement ratio reduces

The aggregate excess reserves beyond statutory requirements in Chinese banking system stood at an average of 10% of
deposit base in the 1990s and the early 2000s (Anderson, 2009), although the ratio gradually fell to 2.3% in 2011, but
compared to banks in the US and Euro-zone countries, it is considered high.


Downloaded by [Purdue University] at 12:49 18 January 2015

1426
the fraction of deposits that banks can use to finance
loans (Romer, 1985; Vargas et al., 2011). In line with

the argument of Thakor (1996), it is assumed that
reserve requirement changes do not greatly affect the
loan return rate, the capital requirement cost and
government securities’ return (opportunity cost) in
the short run. These assumptions are argued to hold
in the Chinese banking market because the reserve
requirements in China are primarily used as a tool to
moderate excess reserves, do not reflect the monetary
policy stance of the PBC (Anderson, 2009) and
appear to have an insignificant effect on the interbank
market rate (Chen et al., 2011). An increase in
deposit cost and unchanged capital requirement cost
reduces the capital requirement cost’s proportion in
the cost of funding for credit lending. In other words,
the cost of funding for credit lending falls relative to
the cost of fund for government securities investments. However, the opportunity cost and loan return
do not change greatly. Therefore, banks are induced
to direct investment funds from government securities to credit lending.
The primary limitation of the credit rationing theory is that it does not consider the liquidity cost. Bank
credit tends be a long-term commitment and costly
to liquidate at short notice (Brunnermeier and
Pedersen, 2009). In contrast, government securities
can easily be converted into cash, which make them
ideal for liquidity contingency. Therefore, banks face
a higher illiquidity risk (which is equivalent to a
higher liquidity cost) when they direct resources to
credit lending instead of government securities. For
this reason, it is argued that the cost of credit lending
includes not only the cost of funding and the opportunity cost as proposed by Thakor (1996), but also
the liquidity cost compared with the cost of investing

in government securities. Because a rise in the statutory reserve requirements drains liquidity from the
banking system and curtails banks’ ability to raise
deposits (Bernanke and Blinder, 1988), the likelihood of a liquidity shortage increases under this
circumstance, and the liquidity cost also increases.
Indeed, the probability of credit rationing increases
because of a higher liquidity cost, and banks tend to
reduce the credit supply in response to the increase in
reserve requirement ratio.
An increase in the reserve requirement ratio leads
to two conflicting effects: although the cost of
2

V. H. T. Nguyen et al.
funding for credit lending falls relative to the cost
of funding for government securities investment
caused by an increase in the deposit cost, the liquidity
cost increases. Because a decrease in the cost of
funding for credit lending augments the credit supply
and an increasing liquidity cost discourages the credit
supply, the effect of an increase in reserve requirements on the credit supply is undetermined.
However, in the presence of IERs, the cost of funding
effect may dominate the liquidity effect. Ceteris paribus, an increase in reserve requirement ratio reduces
the IERs (Agénor et al., 2004). The presence of the
IERs one period after a reserve requirement shock
indicates that the increase in reserve requirements fail
to eliminate unwanted liquidity in the banks.
Therefore, the increase in reserve requirements may
have an insignificant impact on the liquidity of the
banks. If the amount of IERs is positive one period
after an increase in the reserve requirement ratio, the

fall in the cost of funding dominates the increasing
liquidity cost, which results in a greater credit supply.
However, if the amount of IERs is negative one
period after a reserve requirement shock, both the
liquidity effect and the cost of funding effect are at
work in opposing ways, and the impact of reserve
requirement shocks on the credit supply remains
undetermined.
Under the credit rationing theory, banks ration
credit applications because the information asymmetry between banks and potential borrowers may lead
to moral hazards and excessive credit risks to the
banks (Thakor, 1996). In the context of China, the
problem of information asymmetry between banks
and firms is severe because of the poor credit history
of the private sector (Firth et al., 2009). In addition,
the majority of private firms in China are small and
medium enterprises (SMEs) (Allen et al., 2009), and
the asymmetric information that exists with respect to
SMEs arises from the lack of transparency, less information disclosure, an informal accounting system
and weak internal control and governance systems
(Berger and Udell, 2006).
In Thakor’s (1996) model, government securities
do not involve capital back-up (Basel I), although
this pattern does not hold for Basel II2 and Basel III
frameworks, which require banks to take interestrate risks from the securities that they hold into
account as a part of the capital requirement.

The Basel Committee on Banking Supervision, Principles for the Management of Interest Rate Risk, September 1997.

Downloaded by [Purdue University] at 12:49 18 January 2015

Involuntary reserve, credit rationing, China
Furfine (2001) argues that loans are considered
more risky than securities and that the loans therefore require a higher percentage of equity to reflect
their larger risk weight. In other words, the cost of
funding for credit lending always bears an additional capital requirement cost compared to the
cost of funding for securities investment. In addition, the Chinese bond-market capitalization is very
small relative to the credit volume, and the majority
of the bond market consists of central-bank bills
(Hansakul et al., 2009) whose size is too small to
sterilize the excess liquidity in the Chinese banking
market (Conway et al., 2010). Moreover, Chinese
commercial banks are not allowed to engage in trust
investment or stock broking (PRC, 1995, Article
43), which limits the securities investment opportunities of Chinese banks. Therefore, in the Chinese
banking market, the credit rationing theory is analogous to credit lending versus hoarding IERs,
rather than versus investing in securities. As IERs
are not subject to capital requirement regulations,
the capital requirement cost is only present in credit
lending. Regarding the opportunity cost, the PBC
maintains the interest on excess reserves at a fixed
rate below deposit benchmark rate; indeed, there
were only two adjustments in the period 2000–
2011 (Anderson, 2009; Laurens and Maino, 2009;
Ma et al., 2011). For this reason, reserve requirement shocks do not affect the opportunity cost. In
the presence of IERs in the Chinese banking market,
it is argued that an increase in reserve requirement
ratio does not affect the loan return, opportunity cost
or the liquidity cost, but it reduces the relative cost

of funding for credit lending. This is because the
rising deposit cost renders the (additional) capital
requirement cost less significant. Consequently,
Chinese banks tend to expand the credit supply in
response to the increase in reserve requirements. In
light of the above discussion, for banks that have
positive IERs one period after a reserve requirement
shock, it is expected that the credit supply has a
positive relationship with the change in the reserve
requirement ratio. However, for banks with negative
IERs one period after a reserve requirement shock,
the liquidity effect and the cost of fund effect
operate in opposing ways; hence, the impact of
reserve requirement shocks on the credit supply is
undetermined.

1427
III. Methods and Data Analysis
Data and the measure of IER
Banking data covering the period from 2000 to 2011
are collected from Bankscope-Fitch’s International
Bank Database. Only commercial banks whose
data are available for at least three consecutive
years are considered. Other types of banks (i.e.
policy banks, cooperative banks and investment
banks) are not included because they may have
different objectives rather than profitability. The
final sample consists of 95 banks and 552 annual
observations. Monetary policy data are collected
from the PBC website. Furthermore, other macro

data (e.g. national and provincial growth rates of the
real GDP) are collected from the China Securities
Market and Accounting Research database and the
China Statistical Yearbook (the National Bureau of
Statistics of China).
Following the studies of Agénor et al. (2004);
Nguyen and Boateng (2013), we decompose IERs
from precautionary excess reserves. IERs ratio is
the difference between the ratio of actual excess
reserves to deposit and the ratio of the estimated
precautionary excess reserves to deposit. Excess
reserves are defined as the current account holdings of banks, with the central bank that are
beyond the required amount of reserves (Bindseil
et al., 2006). Aikaeli (2011) modifies the precautionary-excess-reserves model by arguing that
banks tend to demand more excess reserves to
buffer the credit risk. Following Agénor
et al. (2004), Aikaeli (2011), and Nguyen and
Boateng (2013), we model the demand for precautionary excess reserves, and the estimation residual is recorded in the form of IER.

ERit ¼ τ þ α1 ERi;tÀ1 þ α2 ðLÞLR þ α3 ðLÞCASH
þ α4 ðLÞYR þ α5 ðLÞARRR þ α6 ðLÞR
þ α7 YEARt þ εit
(1)
where τ is a constant term, εit is a well-behaved
error term and αj ð LÞ are lag polynomials, which are
defined as follows:


V. H. T. Nguyen et al.

1428
αj ¼ 1 þ αj1 L þ . . . þ αjp Lp ; j ! 2

(2)

Table 1. SGMM estimation for precautionary excess
reserves

Downloaded by [Purdue University] at 12:49 18 January 2015

Dependent Variable: ER

ER is the ratio of excess reserves to deposits. ER
is measured as the ratio of the difference between a
bank’s current account balance with the central
bank and the required reserve3 over the total customer deposit. Following Aikaeli (2011), the loanreturn volatility (LR) is used to capture the credit
risk that may trigger deposit withdrawals; LR is
measured as the absolute value of the deviation of
loan interest income from its trend, which is identified by the filter method that was developed by
Hodrick and Prescott (1997). Loan interest income
is the ratio of interest income on loan to total customer deposit. In addition, the Hodrick–Prescott
filter (HP) is a standard method for removing
trend movements in the business cycle literature
(Ravn and Uhlig, 2002). CASH reflects the cashholding preferences of depositors, which are measured based on the volatility of the ratio of vault
cash to total customer deposit by HP filter. YR is the
ratio of real GDP growth rate to its trend (HP filter),
which captures the demand for cash. Moreover,
ARRR and R are the average reserve requirement
ratio set by the PBC within a certain year and the
refinance interest rate, respectively; the latter term

is the rate that the PBC charges when lending to
financial institutions for short-term liquidity support (20-day call loan rate) and reflects the penalty
cost if a bank falls short of the required amount of
reserves. The summary on the statistics and the
results on the unit-root tests for the variables of
precautionary excess reserve estimation are provided in Appendices 1 and 2. The model is estimated by a System Generalized Method of
Moments (SGMM), which was developed by
Arellano and Bond (1991), Arellano and
Bover (1995) and Blundell and Bond (1998). The
number of lags is based on the Aikaike Information
Criteria. The error term εit which is free of unit-root
and serial correlations is collected to index the IER
ratio. The estimation results in Table 1 show that the
demand for precautionary excess reserves has a
significantly positive relationship with the credit
3

Constant
ER (lag1)
LR
CASH
YR
ARRR
R
Number of observations
Number of groups
Number of instruments
Hansen p-value
Second order Arellano–Bond test p-value

−0.006 (0.07)
0.44 (0.27)
0.88** (0.38)
0.22 (0.5)
−0.04 (0.04)
−0.38 (0.32)
3.86** (1.81)
457
95
68
0.927
0.101

Notes: ** denotes statistical significance at 1% level.
Robust SE are reported in parentheses.

risk, which confirms the evidence from the study
of Aikaeli (2011).
SGMM and variable definitions
Following Gambacorta (2005), Gunji and Yuan
(2010), and Nguyen and Boateng (2013), the following dynamic model is used to examine the impact of
reserve requirement shocks on the credit supply in
the presence of IERs in China. Since the sample
covers a relatively short period, only the first lag of
dependent variable is considered, and this is in line
with prior studies (see Altunbaş et al., 2002; Tabak
et al., 2010).
LOANit ¼ αi þ β1 LOANi;tÀ1 þ β2 LIQi;tÀ1
þ β3 SIZEi;tÀ1 þ β4 CAPi;tÀ1
þ β5 IERi;tÀ1 þ β6 NIMi;tÀ1 þ β7 IPtÀ1

þ β8 YtÀ1 þ β9 RRRtÀ1 þ β10 DIERit
þ β11 DIERit  RRRtÀ1 þ εit
(3)
where αi is a constant term and εit is a wellbehaved error term.

The required reserve is measured as the product of the total customer deposit and the reserve requirement ratio for
domestic currency deposits. Because the reserve requirement ratio for foreign currency deposits is smaller than what is
required for Renminbi (RMB) deposits, the total estimated required reserves is slightly higher than the actual value.
However, a comparison with this actual value (where available) shows that the real and estimated required reserves are very
close because foreign currency deposits account for a very small fraction of the total customer deposit.


Involuntary reserve, credit rationing, China

1429

Downloaded by [Purdue University] at 12:49 18 January 2015

The IER is obtained as the residuals from the
estimation of precautionary excess reserves.
Following Gambacorta (2005), we include bank-specific characteristics, namely, liquidity (LIQ), bank
size (SIZE), capitalization (CAP) and net interest
margin (NIM) to control for the bank lending channel. Following Gambacorta (2005), the size (SIZE) is
normalized not just with respect to the mean over the
whole sample period but also with respect to each
single period to remove unwanted trends because
size is measured in nominal terms. As IER is
obtained as regression residuals whose sample
mean equals zero, IER will not be normalized.
Following Gambacorta (2005), other bank-specific

variables (LIQ, CAP and NIM) are normalized
using the mean of the sample as follows:
PN
SIZEit ¼ log Ait À

log Ait
Nt

i¼1

!
T PN
X
Lit
i¼1 Lit =Ait
=T
LIQit ¼
À
Ait
Nt
t¼1

Cit
À
CAPit ¼
Ait

!
T PN
X

i¼1 Cit =Ait
=T
Nt
t¼1

!
T PN
X
i¼1 NIMRit
=T
NIMit ¼ NIMRit À
Nt
t¼1

(4)

(5)

(6)

(7)

where N and T are the numbers of observations
and years, respectively. Moreover, L denotes liquid
assets as defined by BankScope, which includes
cash, government bonds, short-term claims on other
banks (including certificates of deposit) and, where
appropriate, the trading portfolio. C and A refer to
equity (capital) and total assets, respectively.
Because an increase in reserve requirements is considered to be a tax on the banks, if the banks fail to

completely pass this tax onto their borrowers (in the
form of higher lending rates) or depositors (in the
form of lower deposit rates), the banks’ net interest
margin will shrink, thereby reducing the credit supply (Romer, 1985). The model includes net interest
margin (NIM) to take the tax effect into account. In

line with Bankscope’s definition, the net interest
margin ratio (NIMR) is measured as the ratio of net
interest revenue to total earning assets.
Credit supply (LOAN) is defined as the change in
the natural logarithm of gross loan ( Δln(grossloan)),
where grossloan is the total amount of credits that a
bank issues during a particular year. Interest rate
policy (IP) is included to control for the impact of
monetary policy stance on credit supply (see Borio
and Zhu, 2012). Liu et al. (2009) and He and
Wang (2012) argue that the open market operation
rate (the rate at which the central bank sells or buys
government bonds on the open market) in China does
not signal the monetary policy stance of the PBC.
The monetary policy interest rate in China (IP) is
proxied by the change in the one-year deposit benchmark (ceiling) rate (DB) because the policy deposit
ceiling rates are strictly binding and signal a marketclearing equilibrium in China, but the lending benchmark rate is not (Anderson, 2009; Porter and
Xu, 2009). The real GDP growth rate (Y ) is used to
capture the credit demand. Regarding reserve
requirement shock index (RRR), the average of all
of the reserve requirement ratios (ARRR) within a
certain year is taken; then the reserve requirement
ratio shock (RRR) is defined as the change in the
average reserve requirement ratio (ARRR) from the

previous year. Previous studies in the area of monetary policy transmission (e.g. Altunbaş et al., 2002;
Gambacorta, 2005) point out that the credit supply’s
response to the change in the monetary policy rates
rather than the monetary policy rate levels can capture the monetary policy effectiveness. For this reason, the change in reserve requirement ratio (RRR)
instead of the reserve requirement level (ARRR) is
used to reflect the policy shocks to the credit supply
market. DIER is a dummy variable with the value of
1 if IER is positive (IER > 0), and with the value of 0
if IER is negative (IER ≤ 0). The coefficient of the
interaction (β11 ) reflects the difference on credit lending in response to reserve requirement shocks of the
two groups, that is banks with positive IER versus
banks with negative IER one period after the shock.
A summary of the variable statistics is presented in
Table 2. The IER ranges from −22% to 33% of the
total deposit and is positive in 43% of the observations. During the sample period, the average reserve
requirement ratio (ARRR) has a mean of 12.1% and
reaches a peak of 20.36% for the six largest banks
and 18.36% for the other smaller banks (the PBC has


V. H. T. Nguyen et al.

1430

Downloaded by [Purdue University] at 12:49 18 January 2015

Table 2. Summary statistics for reserve requirement impact estimations variables
Variable

Mean

SD

Skewness

Kurtosis

Min

Max

Jarque-bera

L/A
LIQ
C/A
CAP
SIZE
IER
DB
IP
Y
LOAN
NIMR
NIM
ARRR
RRR

0.25
−0.009

0.085
−0.005
0
0
0.026
0.001
0.099
0.208
2.58
−0.35
0.121
0.015

0.125
0.125
0.111
0.111
2.058
0.046
0.005
0.005
0.010
0.191
0.82
0.82
0.04
0.019

1.923
1.923

3.84
3.84
0.522
1.35
0.877
−0.669
−0.136
1.670
0.642
0.642
0.033
−0.353

5.427
5.427
17.135
17.135
0.006
11.712
0.146
1.876
−0.494
14.029
1.55
1.55
−1.336
−0.825

0.035
−0.223

−0.137
−0.229
−4.724
−0.219
0.0198
−0.0234
0.071
−0.559
0.195
−2.739
0.06
−0.0217

0.893
0.635
0.872
0.78
5.202
0.328
0.0459
0.0095
0.114
1.761
5.652
2.718
0.2036
0.042

1064*
1064*

8486*
8486*
26.53*
2651*
99.91*
169.3*
10.44*
4104*
95.31*
95.31*
37.6*
24.89*

Note: *denotes the rejection of normal distribution at the 1% significance level.

maintained a two-tier reserve requirement system
since 2008). From 2000 to 2002, the PBC kept the
reserve requirement ratio constant. In contrast, from
2002 to 2011, the reserve requirement ratio was
increased every year except 2009.
Table 3 presents the panel unit-root tests results for
all variables. Augmented Dickey–Fuller and
Phillips–Perron unit root tests (Fisher-type tests
defined by Maddala and Wu, 1999; and
Choi, 2001) for panel data indicate that all variables
are stationary.
Because OLS is biased in dynamic models,
‘System’ GMM estimator is used. Arellano and
Bover (1995) and Blundell and Bond (1998) developed SGMM based on Arellano and Bond (1991)
‘difference’ GMM (DGMM). SGMM is able to deal

with the endogeneity and fixed effects in dynamic
Table 3. Unit root tests for reserve requirement impact
estimations variables
Variable

Augmented Dickey–Fuller

Phillips–Perron

IP
Y
LOAN
NIM
IER
LIQ
CAP
SIZE
RRR

167.9752*
644.8130*
604.7481*
600.7662*
575.7617*
311.8068*
244.7292*
252.3623*
240.9465*

267.9752*

334.8188*
604.7481*
600.7662*
575.7617*
311.8068*
244.7292*
262.1405*
265.6243*

Note: *denotes the rejection of the unit root hypothesis at
the 1% significance level.

models (Arellano and Bover, 1995); furthermore, it
can overcome the weakness of ‘difference’ GMM,
which is inconsistent in the estimations on unbalanced panel data (Roodman, 2006). The lags of
regressors are used as instruments. IP, RRR and the
interaction are treated as endogenous variables. Y is
treated as an exogenous variable. Other variables are
considered to be predetermined. SGMM is implemented by comment xtabond2 in STATA. The optimal model is selected based on the criteria suggested
by Arellano and Bond (1991) and Roodman
(Roodman, 2006, 2009) in Appendix 3.

IV. Estimations Results and Discussion
The results from the estimations are reported in
Table 4 (estimation 1), and the residuals are free of
unit-root and serial correlation. Regarding the control
variables, IER significantly increases credit supply.
Bank size (SIZE) and capital (CAP) have positive
impacts, while liquidity (LIQ) has a negative impact
on credit supply at significant level of 10%. Net

interest margin (NIM), monetary policy interest rate
(IP) and GDP (Y) do not statistically affect credit
supply.
The IER dummy variable (DIER) is not statistically significant, indicating that there is no difference
in credit supply between banks with positive and
negative IERs, ceteris paribus. The impact of reserve
requirement shock on credit supply is measured as
follows:


Involuntary reserve, credit rationing, China

1431

Downloaded by [Purdue University] at 12:49 18 January 2015

Table 4. SGMM estimations for reserve requirement impact on credit supply (LOAN)
Dependent variable: LOAN

(1)

IER > 0 (2)

Constant
LOAN (lag1)
RRR
DIER
RRR × DIER
IER
LIQ

SIZE
CAP
NIM
IP
Y
Number of observations
Number of groups
Number of instruments
Hansen p-value
Second order Arellano–Bond test p-value

0.04 (0.7)
0.17** (0.08)
−0.92 (1.36)
−0.002 (0.05)
4.09** (2.05)
0.55** (0.26)
−0.41* (0.24)
0.04* (0.02)
0.66* (0.36)
−0.003 (0.04)
−7.94 (6.28)
1.03 (3.48)
331
89
80
0.798
0.851

IER < 0 (3)

−0.34 (0.47)
0.62*** (0.16)
2.41** (1.09)

−0.04 (1.03)
0.08 (0.25)
−1.37 (2.14)

0.64** (0.28)
−0.2 (0.44)
−0.04 (0.04)
0.32 (0.72)
−0.02 (0.05)
−14.54** (8.07)
4.04 4.51
151
69
37
0.901
0.278

1.22* (0.73)
−0.38 (0.66)
−0.03 (0.07)
0.82 (1.09)
0.12 (0.11
−7.55 (16.88)
2.94 (10.59)
186

71
23
0.752
0.616

Notes: ***, ** and * denote statistical significance at the 1%, 5% and 10% significance levels, respectively.
Robust SE are reported in parentheses.

@LOAN
¼ β9 þ β11 Â DIER
@RRR

(8)

For banks with negative IER one period after
reserve requirement shocks (the value of DIER
equals 0), the impact of reserve requirement shocks
on credit supply is reflected on β9 , which is negative
and not statistically significant. This supports the
argument that the liquidity effect and the cost of
fund effect operate in opposing ways in response to
reserve requirement shocks for banks with negative
IERs, and hence, the impact of reserve requirement
shocks on the credit supply is undetermined.
For banks with positive IERs one period after
reserve requirement shocks (the value of DIER
equals 1), the impact of reserve requirement shocks
on credit supply is
@LOAN
¼ β9 þ β11 ¼ À0:92 þ 4:09 ¼ 3:17

@RRR
The result shows that the coefficient of the interaction β11 is positive, statistically significant and
much greater than β9 . The sum of β9 and β11 is
positive, indicating that banks with positive IERs
one period after reserve requirement shocks tend to
increase credit supply in response to increases in
reserve requirement ratio. The model is further

estimated separately for two groups, that is banks
with negative IERs (IERit ≤ 0) and positive IERs
(IERit > 0) one period after reserve requirement
shocks (RRRt−1) without the IER dummy and the
interaction. The results in Table 4 (estimations 2 and
3) show that the impact of reserve requirement
shock (β9 ) on credit supply is positive and statistically significant for banks with positive IERs but
not significant for banks with negative IERs. This
evidence supports the following argument: if an
increase in the reserve requirement ratio fails to
eliminate IERs completely (i.e. if positive IERs
remain one period after the hike in reserve requirement ratio), banks tend to expand their credit supply
in response to this increase in reserve requirement
ratio. This finding contradicts the evidence from
prior studies, which report the negative relationship
between the reserve requirement ratio and the credit
supply (e.g. see Takeda et al., 2005; Cargill and
Mayer, 2006; Mora, 2009). One possible reason
for the difference is that the prior studies do not
consider IERs and the cost of funding effect.
These studies therefore overestimate the liquidity
effect and deduce that there is a negative relationship between the reserve requirement ratio and the

credit supply. However, this finding supports the
study of Qin et al. (2005) who find that an increase
in reserve requirement ratio generates a small rise in
GDP growth rate in China.


V. H. T. Nguyen et al.

1432
V. Additional Analysis and Robustness
Tests
The robustness tests are reported in Table 5, and the
tests are summarized in Table 6. The PBC employs a
loan ceiling as a monetary policy tool to moderate the

credit supply; its primary target is the four stateowned commercial banks (SOCB) (Geiger, 2008).
These loan limits may make the four state-owned
commercial banks less responsive to reserve requirement shocks. To address the effect of loan limits, we
exclude the four state-owned banks from the sample.

Table 5. Estimation results for additional analysis and Robustness tests
Dependent variable: LOAN

(4) (without SOCB) (5)

(6)

Constant

−0.41

(0.8)
−0.02
(0.25)
−0.9
(1.05)
−0.02
(0.05)
3.14**
(1.41)
0.94***
(0.34)
−0.1
(0.29)
0.01
(0.02)
0.99
(0.99)
0.06
(0.06)
−15.97
(12.4)
6.53
(8.03)

−0.25
−0.35
−0.44
(0.38)
(0.25)
(0.61)

0.19**
0.11
0.21
(0.09)
(0.09)
(0.2)
−1.04
−0.24
−0.57
(1.53)
(1.01)
(1.34)
0.02
−0.003
0.02
(0.05)
(0.04)
(0.05)
4.62**
3.15**
4.42**
(2.32)
(1.48)
(2.27)
0.62**
0.54**
0.92**
(0.25)
(0.26)
(0.44)

−0.06
−0.26
−0.08
(0.28)
(0.19)
(0.22)
0.05**
0.06*
−0.02
(0.03)
(0.03)
(0.02)
0.51
0.75
0.41
(0.36)
(0.51)
(0.53)
0.009
−0.005
−0.57
(0.05)
(0.04)
(0.07)
−12.69** −13.96*** −17.46
(6.34)
(3.71)
(10.7)
2.94
6.09

(3.36)
(6.32)

LOAN (lag1)

Downloaded by [Purdue University] at 12:49 18 January 2015

RRR
DIER
RRR × DIER
IER
LIQ
SIZE
CAP
NIM
IP
Y
LP

0.09
(0.41)
0.17**
(0.08)
−1.13
(1.22)
0.006
(0.04)
3.97**
(1.91)
0.61**

(0.26)
−0.42*
(0.25)
0.05**
(0.02)
0.64
(0.26)
0.004
(0.04)
0.5
(3.91)
−5.32
(5.86)

IOR

(7)

(8)

(9)

(10)

0.39
(0.37)
0.17*
(0.09)
0.28
(1.08)

0.06
(0.05)
2.69*
(1.55)
0.52**
(0.24)
0.13
(0.3)
0.06**
(0.03)
0.72**.
.(0.35)
0.03
(0.04)
−4.37
(6.12)
−2.08
(3.45)

0.1
(0.38)
0.21**
(0.09)
−1.09
(1.75)
−0.01
(0.67)
4.84*
(2.91)
0.65**

(0.26)
−0.44*
(0.26)
0.04*
(0.02)
0.59
(0.39)
−0.01
(0.04)
−7.4
(6.33)
0.36
(3.49)

10.38*
(5.56)

RGDP

4.3**
(2.03)

OWNERSHIP_NSOB
OWNERSHIP_F

−0.01
(0.09)
−0.2
(0.16)

ACCOUNTING

−0.09***
(0.03)

M&A
Number of observations
306
Number of groups
85
Number of instruments
38
Hansen p-value
0.661
Second order Arellano–Bond test
0.708
p-value

331
89
82
0.824
0.988

331
89
67
0.621
0.698

331
89
82
0.862
0.486

331
89
54
0.469
0.794

331
89
83
0.882
0.896

Notes: ***, ** and * denote statistical significance at the 1%, 5% and 10% significance levels, respectively.
Robust SE are reported in parentheses.

−0.08
(0.06)
331
89
62
0.496
0.765

Involuntary reserve, credit rationing, China

1433

Table 6. Summary on additional analysis and Robustness tests
Issue

Test performed

Specification

Downloaded by [Purdue University] at 12:49 18 January 2015

Loan ceilings set for four SGMM estimation for The four state-owned banks are
excluded from the sample.
the sample without
state-owned commercial
the four state-owned
banks
commercial banks

Finding
No significant difference in the
results compared with the main
estimation, which includes the
four state-owned commercial
banks in the sample.
No significant difference in the
results compared with the main
estimation, which uses deposit

benchmark rate as an index for
monetary policy.
No significant difference in the
results compared with the main
estimation, which does not
include without interest on
excess reserve.
No significant difference in the
results compared with the main
estimation where national real
GDP growth rate is used for all
banks.

Alternative index of
monetary policy

SGMM estimation
including lending
benchmark rate

Lending benchmark rate is used
instead of deposit benchmark
rate to index monetary policy

Excess reserve
remuneration

SGMM estimation
including interest
on excess reserves

Interest on excess reserve is used
to capture the effect of excess
reserve remuneration.

Provincial operation

SGMM estimation
with provincial
GDP growth rate

Ownership structure and
operational objective

SGMM estimations
with ownership
dummies

Data consistency:
introduction of two-tier
reserve requirement
system and changes in
Chinese accounting
standards in 2008
Merger and Acquisition
(M&A)

SGMM estimations
with accounting
dummies

National real GDP growth rate is
applied for State-owned
commercial banks, joint stock
commercial banks and foreign
banks. Provincial real GDP
growth rate is applied for city
commercial banks and rural
commercial banks.
No significant difference in the
Include Ownership Dummies
results compared with the main
such that 1, 2 and 3 are
estimation, which does not
dummies for state-owned,
include ownership dummies.
nonstate-owned and foreignowned banks, respectively
No significant difference in the
Include Accounting Dummies
results compared with the main
such that 1 and 2 are dummies
estimation which does not
for periods prior to and after
include accounting dummies.
2008, respectively

SGMM estimations
controlled for
M&A

Include M&A dummies with
value of 1 for merger and
acquisition event and 0
otherwise

However, the estimations without the four stateowned commercial banks have similar results to the
estimations that include these four banks. To further
check for the robustness, the model is estimated
using lending benchmark (LP) rate as an index for
monetary policy interest rate as contrast to the
deposit benchmark rate in the main estimation. The
robustness test’s results show no significant difference to that of the main estimation. In addition, as
excess reserve remuneration may affect credit supply,
the robustness test including interest on excess
reserves (IOR) is carried out, and there is no

No significant difference in the
results compared with the main
estimation which does not
include M&A dummies.

significant difference in the results compared to the
main estimation, which does not include IOR.
Xu (2011) points out that city commercial banks
and rural commercial banks in China tend to operate
on a provincial scale rather than a national scale. To
take into account the effect of credit demand, the
provincial GDP growth rate is used for both city
commercial banks and rural commercial banks,
whereas the national GDP growth rate is applied to

other types of banks. The results show no significant
difference from the main estimation, where the
national GDP growth rate is applied to all banks.


Downloaded by [Purdue University] at 12:49 18 January 2015

1434
Moreover, it is argued that the Chinese monetary
policy is more enforceable against Chinese stateowned banks than nonstate-owned banks
(Geiger, 2008). It has also been suggested that foreign banks are less prone to liquidity effect of reserve
requirement shocks because foreign banks can
receive liquidity support from their home banks
(Tabak et al., 2010; Ahtik, 2012). To ensure that the
effects of reserve requirements on the credit lending
behaviours of banks that have and lack IERs are
robust across different ownership structures, we
include ownership dummies, that is state-owned,
nonstate-owned (OWNERSHIP_NSOB) and foreign
ownership (OWNERSHIP_F) in the estimations.
The results are more or less identical to the main
estimation, which has no ownership dummies.
Since 2008, the PBC has adopted a two-tier reserve
requirements system in which the reserve requirement
ratio for the six largest commercial banks is 2% higher
than that applied to other banks (Ma et al., 2011). In
addition, the new accounting standards introduced in
China in 2008 may cause data inconsistency. To take
into account the effects from these events, the robustness test is conducted by including an accounting
dummy (ACCOUNTING) to capture the difference

between the two periods (prior to and after 2008). The
results show no significant difference from the main
estimation, which does not include an accounting
dummy. Moreover, we control for the effects of mergers and acquisitions (M&A) on the reporting of
financial statements by including an M&A dummy;
the results are essentially identical to the main estimations which lacks an M&A dummy.

VI. Summary and Policy Implications
This study examines the impact of the changes in the
reserve requirement ratio on the credit lending behaviour of banks in China, where IERs in the banking
system are high. This study argues that reserve
requirement shocks impose not only a liquidity effect
but also the cost of funding effect on the credit
supply. An increase in the reserve requirement ratio
appears to increase the liquidity cost but reduce the
cost of fund for credit lending relative to the cost of
funding for securities investments. The findings of
this study indicate that IERs render the liquidity

V. H. T. Nguyen et al.
effects of reserve requirement shocks insignificant.
Therefore, in the presence of IERs, the cost of funding effect dominates the liquidity effect. The study
finds that banks with positive IERs one period after a
reserve requirement shock tend to increase the credit
supply in response to an increase in the reserve
requirement ratio. However, the impact of reserve
requirement shocks on the credit supply appears to
be insignificant for banks with negative IERs one
period after a reserve requirement shock.
The PBC actively uses reserve requirements as a

monetary policy tool to reduce excess liquidity and
moderate the credit supply in the Chinese banking
market (Anderson, 2009; Conway et al., 2010; Ma
et al., 2011). However, if an increase in the reserve
requirement ratio fails to completely eliminate IERs,
this increase in reserve requirements unexpectedly
tends to induce Chinese banks to increase the credit
supply, which makes the reserve requirement instrument not only ineffective but also counterproductive.
For this reason, it is suggested that the PBC should
take into account the trade-off effect of the reserve
requirements between the reduction of excess liquidity
and the expansion of credit. In addition, to discourage
lending, the PBC should consider increasing the interest rate on excess reserves. At a higher interest rate for
excess reserves’ remuneration, the opportunity cost on
the credit supply increases and banks may therefore
increase credit rationing and reduce the credit supply.
This study has examined an under-researched area
with respect to the impacts of reserve requirements
on the credit supply in an emerging market economy,
where the level IERs consistently appears to be high.
The findings are interesting but preliminary because
of the sample size. When quarterly data becomes
available, further studies appear to be warranted to
better capture the instantaneous credit supply
response to reserve requirement shocks.

Acknowledgement
The authors would like to thank the Journal’s referees
for helpful comments on earlier drafts of the article.

Disclosure statement
No potential conflict of interest was reported by the
authors.


Involuntary reserve, credit rationing, China

Downloaded by [Purdue University] at 12:49 18 January 2015

References
Agénor, P.-R., Aizenman, J. and Hoffmaister, A. W.
(2004) The credit crunch in East Asia: what can
bank excess liquid assets tell us?, Journal of
International Money and Finance, 23, 27–49.
doi:10.1016/j.jimonfin.2003.08.008
Ahtik, M. (2012) Bank lending channel in Slovenia: panel
data analysis, Prague Economic Papers, 1, 50–68.
Aikaeli, J. (2011) Determinants of excess liquidity in
Tanzanian commercial banks, African Finance
Journal, 13, 47–63.
Allen, F., Qian, J., Qian, M. et al. (2009) A review of
China’s financial system and initiatives for the future,
in China’s Emerging Financial Markets – Challenges
and Opportunities, Barth, J. R., Tatom, J. A. and
Yago, G. (Eds), Milken Institute, Santa Monica, CA.
Altunbaş, Y., Fazylov, O. and Molyneux, P. (2002)
Evidence on the bank lending channel in Europe,
Journal of Banking and Finance, 26, 2093–110.
doi:10.1016/S0378-4266(02)00201-7
Anderson, J. (2009) The China monetary policy handbook, in China’s Emerging Financial Markets –

Challenges and Opportunities, Barth, J. R., Tatom,
J. A. and Yago, G. (Eds), Milken Institute, Santa
Monica, CA.
Arellano, M. and Bond, S. (1991) Some tests of specification for panel data: Monte Carlo evidence and an
application to employment equations, The Review of
Economic Studies, 58, 277–97. doi:10.2307/2297968
Arellano, M. and Bover, O. (1995) Another look at the
instrumental variable estimation of error-components
models, Journal of Econometrics, 68, 29–51.
doi:10.1016/0304-4076(94)01642-D
Berger, A. and Udell, G. (2006) A more complete conceptual framework for SME finance, Journal of
Banking and Finance, 30, 2945–66. doi:10.1016/j.
jbankfin.2006.05.008
Bernanke, B. S. and Blinder, A. S. (1988) Credit, money,
and aggregate demand, American Economic Review,
78, 435–9.
Bindseil, U., Camba-Mendez, G., Hirsch, A. et al. (2006)
Excess reserves and the implementation of monetary
policy of the ECB, Journal of Policy Modeling, 28,
491–510. doi:10.1016/j.jpolmod.2006.02.006
Blundell, R. and Bond, S. (1998) Initial conditions and
moment restrictions in dynamic panel data models,
Journal of Econometrics, 87, 115–43. doi:10.1016/
S0304-4076(98)00009-8
Borio, C. and Zhu, H. (2012) Capital regulation, risktaking and monetary policy: a missing link in the
transmission mechanism?, Journal of Financial
Stability, 8, 236–51.
Brunnermeier, M. K. and Pedersen, L. H. (2009) Market
liquidity and funding liquidity, Review of Financial
Studies, 22, 2201–38. doi:10.1093/rfs/hhn098

Cargill, T. F. and Mayer, T. (2006) The effect of changes
in reserve requirements during the 1930s: the evidence from nonmember banks, The Journal of

1435
Economic History, 66, 417–32. doi:10.1017/
S0022050706000179
Chen, H., Chen, Q. and Gerlach, S. (2011) The implementation of monetary policy in China: the interbank
market and bank lending, HKIMR Working Papers,
No. 26, HKIMR, Hong Kong.
Choi, I. (2001) Unit root tests for panel data, Journal of
International Money and Finance, 20, 249–72.
doi:10.1016/S0261-5606(00)00048-6
Conway, P., Herd, R. and Chalaux, T. (2010) Reforming
China’s monetary policy framework to meet domestic objectives, OECD Economics Department
Working Papers, No. 822, OECD Library, Paris.
Firth, M., Lin, C., Liu, P. et al. (2009) Inside the black box:
bank credit allocation in China’s private sector,
Journal of Banking and Finance, 33, 1144–55.
doi:10.1016/j.jbankfin.2008.12.008
Furfine, C. (2001) Bank Portfolio allocation: the impact
of capital requirements, regulatory monitoring,
and economic conditions, Journal of Financial
Services Research, 20, 33–56. doi:10.1023/
A:1011147609099
Gambacorta, L. (2005) Inside the bank lending channel,
European Economic Review, 49, 1737–59.
doi:10.1016/j.euroecorev.2004.05.004
Geiger, M. (2008) Instruments of monetary policy in
China and their effectiveness: 1994-2006, in United
Nations Conference on Trade and Development,

No. 187, UNCTAD, New York.
Gunji, H. and Yuan, Y. (2010) Bank profitability and the
bank lending channel: evidence from China, Journal
of Asian Economics, 21, 129–41. doi:10.1016/j.
asieco.2009.12.001
Hansakul, S., Dyck, S. and Kern, S. (2009) China’s
Financial Markets – A Future Global Force?,
Deutsche Bank Research, Berlin.
He, D. and Wang, H. (2012) Dual-track interest rates and
the conduct of monetary policy in China, China
Economic Review, 23, 928–47. doi:10.1016/j.
chieco.2012.04.013
Hodrick, R. and Prescott, E. C. (1997) Postwar U.S. business cycles: an empirical investigation, Journal of
Money, Credit and Banking, 29, 1–16. doi:10.2307/
2953682
Laurens, B. J. and Maino, R. (2009) Monetary policy
implementation in China: past, present and prospects, in China’s Emerging Financial Markets –
Challenges and Opportunities, Barth, J. R., Tatom,
J. A. and Yago, G. (Eds), Milken Institute, Santa
Monica, CA.
Liu, L.-G. and Zhang, W. (2010) A new Keynesian model
for analysing monetary policy in Mainland China,
Journal of Asian Economics, 21, 540–51.
doi:10.1016/j.asieco.2010.07.004
Liu, M.-H., Margaritis, D. and Tourani-Rad, A. (2009)
Monetary policy and interest rate rigidity in China,
Applied Financial Economics, 19, 647–57.
doi:10.1080/09603100801998576

V. H. T. Nguyen et al.

Downloaded by [Purdue University] at 12:49 18 January 2015

1436
Ma, G., Yan, X. and Liu, X. (2011) China’s evolving
reserve requirements, BIS Working Papers, No.
360, BIS Press & Communication, Basel.
Maddala, G. S. and Wu, S. (1999) A comparative study of
unit root tests with panel data and a new simple test,
Oxford Bulletin of Economics and Statistics, 61,
631–52. doi:10.1111/1468-0084.61.s1.13
Mora, N. (2009) Reason for reserve? Reserve requirements,
dollarization, and credit, in 30th Annual Meeting of
The Middle East Economic Association Papers,
Conference Proceedings, Atlanta, GA.
Nguyen, V. H. T. and Boateng, A. (2013) The impact of
excess reserves beyond precautionary levels on bank
lending channels in China, Journal of International
Financial Markets, Institutions and Money, 26, 358–
77. doi:10.1016/j.intfin.2013.07.002
Porter, N. and Xu, T. (2009) What drives China’s interbank market?, IMF Working Paper, WP/09/189,
IMF, Washington, DC.
PRC. (1995). Law of the People’s Republic of China on
Commercial Banks. Available at na.
org.cn/english/DAT/214824.htm
(accessed
24
December 2014).
Qin, D., Quising, P., He, X. et al. (2005) Modeling monetary transmission and policy in China, Journal of

Policy Modeling, 27, 157–75. doi:10.1016/j.
jpolmod.2004.12.005
Ravn, M. O. and Uhlig, H. (2002) On adjusting the
Hodrick-Prescott filter for the frequency of observations, Review of Economics and Statistics, 84, 371–6.
doi:10.1162/003465302317411604
Romer, D. (1985) Financial intermediation, reserve
requirements, and inside money: a general equili-

brium analysis, Journal of Monetary Economics,
16, 175–94. doi:10.1016/0304-3932(85)90029-7
Roodman, D. (2006). How to do xtabond2: an introduction to “difference” and “system” GMM in Stata,
Center for Global Development Working Paper No.
103, CGD Press Center, Washington, DC.
Roodman, D. (2009) Practitioners’ corner – A note on the
theme of too many instruments, Oxford Bulletin of
Economics and Statistics, 71, 135–58. doi:10.1111/
j.1468-0084.2008.00542.x
Tabak, B., Laizy, M. and Cajueiro, D. (2010) Financial
stability and monetary policy – The case of Brazil,
Central Bank of Brazil, Working Paper Series No.
217, Research Department, Banco Central do Brasil,
Brasília.
Takeda, T., Rocha, F. and Nakane, M. (2005) The reaction
of bank lending to monetary policy in Brazil, Revista
Brasileira de Economia, 59, 107–26. doi:10.1590/
S0034-71402005000100005
Thakor, A. V. (1996) Capital requirements, monetary policy, and aggregate bank lending: theory and empirical evidence, The Journal of Finance, 51, 279–324.
doi:10.1111/j.1540-6261.1996.tb05210.x
Vargas, H., Betancourt, Y., Varela, C. et al. (2011) Effects of
reserve requirements in an inflation targeting regime:

the case of Colombia, The Global Crisis and Financial
Intermediation in Emerging Market Economies, 54,
133–69.
Xu, Y. (2011) Towards a more accurate measure of foreign bank entry and its impact on domestic banking
performance: the case of China, Journal of Banking
and Finance, 35, 886–901. doi:10.1016/j.
jbankfin.2010.10.011

Appendix 1. Summary statistics for precautionary excess reserve estimation
Variable

Mean

SD

Skewness

Kurtosis

Min

Max

Jarque-Bera

ER
LR
CASH
YR
ARRR

R

0.052
0.009
0
1.004
0.121
3.113

0.056
0.023
0.008
0.061
0.04
0.276

1.836
6.496
9.652
−0.305
0.033
−0.788

6.362
50.994
197.945
−0.016
−1.336
−1.522

−0.1
0
−0.05
0.882
0.06
2.7

0.402
0.257
0.139
1.095
0.204
3.33

1309*
62000*
980000*
0.2521
37.6*
1.947

Note: * denotes the rejection of normal distribution at 1% significance level.


Involuntary reserve, credit rationing, China

1437

Appendix 2. Unit root tests for variables of precautionary excess reserve estimation
Variables

Augmented Dickey–Fuller(1)

Phillips–Perron(1)

KPSS(2)

ER
LR
CASH
YR
ARRR
R

569.5294*
433.4773*
515.2204*
−3.462*
0.459
−3.117**

569.5294*
433.4773*
515.2204*
−3.746*
1.122
−2.221

0.182*
0.0987***

Notes: 1. * and ** denote the rejection of the unit root hypothesis at 1% and 5% significance
levels, respectively.
2. * and *** denote the fail to reject the trend stationarity at 1% and 10% significance levels,
respectively.

Downloaded by [Purdue University] at 12:49 18 January 2015

Appendix 3. xtabond2 model selection criteria
Criteria

Requirements description

F-test
Arellano–Bond test

Reject the null hypothesis that independent variables are jointly equal to zero
First-order serial correlation but no second-order serial correlation in the residuals (Arellano and
Bond, 1991)
Sargan statistic is biased in one-step estimator with ‘Robust’ option (Roodman, 2006). Therefore,
Sargan Test is not considered.

Sargan Test

● p-value ≥0.25 (Roodman, 2009)
Hansen J-statistic
Difference-in-Hansen ● p-value of 1 is the sign of inappropriate model (Roodman, 2009)
Steady state
The estimated coefficient on the lagged dependent variable should have a value less than (absolute)
unity (Roodman, 2009)

Number of
The number of instruments should not exceed the number of groups (i.e. number of banks)
instruments
(Roodman, 2009)
Optimal instruments Roodman (Roodman, 2006, Roodman, 2009) suggests reporting how the optimal number of
instruments. The standard treatment on lag-limits is used, such that lag-limits start from lag2 for
endogenous variable (and from lag1 for exogenous and predetermined) to the most available lag.
The ‘collapse’ option is used to keep the number of instruments within Stata’s size limit. A
number of other regressions are estimated by adjusting the upper and lower lag-limits. The
regression which satisfies all the criteria listed above and has highest p-value of Hansen J test is
selected as the optimal regression.

Note: Compiled by the authors based on Roodman (2006); Roodman (2009); Arellano and Bond (1991).