Credit Access in
India
Abhijit V. Banerjee
How well is capital
allocated?
Prima Facie Evidence
Figure 1. Average Firm Size in India and Comparator
Countries in 1990
10.00
(in US$
millions
establis
in 1990
hment
added
Value
) per
9.00
8.00
7.00
6.00
5.00
4.00
3.00
2.00
1.00
0.00
Textiles Iron and steel Transport Food products Machinery Machinery, Industrial Other Other non- All industries
equipment except electric electric chemicals chemicals metallic
mineral
products
Top nine ISIC 3 digit industries by value added for India in 1990
India Comparator countries
More…
TFPR dispersion within 4 digit industries (from
Hsieh and Klenow)
U.S. China India
90th/10th 1.9 5.6 5.7
75th/25th 1.3 2.5 2.4
A natural experiment on credit access:
Priority Sector Lending in India
Based on Banerjee-Duflo (2004)
Banks need to lend 40% of their portfolio to the
priority sector, including SSI.
Changes in the priority sector rules:
January 1998: inclusion in the SSI sector of firms with
investment in plant and machinery between 0.65 and 3
crores rupees.
Early 2000: firms with ipm between 3 crores rupees
and 1 crore rupees taken away from the priority sector.
Results
We firms level data from 1 very well-
regarded public bank.
We estimates the effect on credit, sales,
and profits, using firms that were always in
the priority sector as controls.
The firms that were included in the priority
sector and then excluded grew much
faster in all 3 measures when included,
and fell behind on all 3 when excluded.
Average change in limit
Year
1997 1998 1999 2000 2001 2002
small 0.111 0.076 0.076 0.048 0.085 0.080
big 0.054 0.113 0.085 0.018 0.003 0.041
difference -0.057 0.036 0.009 -0.030 -0.082 -0.039
Effect on sales and profit
Dependent variables
Log(sales) t+1 Log(profit) t+1 Log(sales) t+1 Log(profit) t+1
-log(sales)t -log(profit) t -log(sales)t -log(profit) t
1996-1999 1998-2001
post -0.007 -0.078 0.011 -0.118
(.048) (.082) (.052) (.152)
big -0.121 -0.255 0.109 0.014
(.074) (.288) (.065) (.163)
post*big 0.164 0.289 -0.193 -0.155
(.08) (.33) (.106) (.236)
452 389 454 376
Implications
1 percent increase in credit generates 1 percent increase in sales
and 3 percent increase in profits
Even after taking default rates (NPA) into account, the implied
marginal product of capital is close to 100%! Average rates of
NPA do not explain under-lending.
Evidence of systemic failure: Neither private lenders nor public
banks are supplying these largish firms with the credit they need.
Direct evidence from interest rates
There is a large body evidence showing that many
smaller firms are also extremely short of credit.
In Banerjee (2002) I review evidence showing that small
firms often pay interest rates of over 50% of which at
most 5% is explained by default.
This tells us that their marginal product must be above
50% whereas the cost of capital is of the order of 5-6% at
most.
On the other hand, from the ICOR for India (4.5), the
average marginal product in India cannot be more than
22%.
So a lot of firms must be earning a lot less than 22%.
Evidence from within industry
allocation
Evidence from the knitted garment industry in Tirupur.
The firms that are associated with a cash-rich community
start out almost three times larger than those started by
other people.
These firms very soon fall behind in terms of output, but
continue to have significantly more fixed capital.
What are lenders
(not) doing?
Lending rules for public sector
banks
Maximum Permissible Bank finance: Since the Nayak
Committee, banks can set their own rule (turnover based,
or based on working capital gap)
For example in the bank we study, MPBF is the
maximum of turnover based limit and the limit based on
the working capital gap
Comments:
Rules set for limited growth.
Profitability does not enter in the official rule.
Inventories are not a very good collateral in practice
However the problem goes
beyond the rules
Banerjee and Duflo (2001) look at the actual lending decisions
of a bank, and compare it with the actual limit.
Main results:
The lending limit changes very infrequently (in 64% of the case, it does
not change) despite growth
Increases are not very responsive to firms’ characteristics and
performance.
Lending in smaller than MPBF in 68% of the cases.
Systematic deviation from the rule in the direction of inertia.
Inertia in Lending
1997 1998 1999
(1) (2) (3)
proportions of cases in which
Granted limit remained the same 0.66 0.64 0.65
Limit was attained by the borrower 0.80 0.69 0.72
Granted limit from banking system remained the same 0.66 0.63 0.63
Maximum authorized limit has increased 0.63 0.74 0.73
Predicted sales have increased 0.72 0.67 0.73
Granted limit <maximum authorized limit 0.60 0.63 0.60
Granted limit <0.20*predicted sales 0.85 0.85 0.80
Inertia in Lending Decisions
proportion of cases
where limit was not changed
A- PAST UTILIZATION C. PROFIT OVER SALES
REACHED LIMIT INCREASED
Yes 0.60 0.65
No 0.66 0.61
Difference -0.06 0.04
(.064) (.048)
B-CURRENT SALES D. CURRENT RATIO
INCREASED INCREASED
Yes 0.62 0.62
No 0.68 0.65
Difference -0.06 -0.03
(.048) (.044)
Why are they not
lending?
Understanding why banks do not
lend?
Lack of positive incentives
Fear of lending
Lending to the government and the easy
life
The risk of (marginal) default
Fear of Lending
Employees of public bank are subject to anti-
corruption legislation: widespread concerns about the
legal proceedings.
No incentive to lend more: it is easier to do nothing.
Using monthly data on lending combined with public
data on CVC investigation, we examine whether there
is a decrease in lending in a bank following CVC
activity.
Results
Following vigilance activity in a particular bank, total
lending by the bank drops by 3-5%, compared to other
banks and stay low for several years.
This understates the overall impact since there might be
some anticipatory reaction as well (we have only data on
the date at which the CVC advice was given).
This could imply a sizeable reduction in the credit supply
in the economy.
What is going on?
The attraction of easy life
The combination of high interest rates on government
borrowing and a boom in consumer finance as the economy
transitions to credit financing of durables reduces the pressure
to lend to industry.
Test: Are banks in slow growing states more responsive
to variation in government interest rates then banks in
fast growing states ?
Data: yearly data on C/D ratio for 45 banks
Results: in high growth environment, banks are less
elastic to the spread in their lending decisions
Results
Synthetic Growth Index
Time Period 1985-2000 1992-2000
(3) (4)
Growth 2.195 2.634
(0.970) (1.165)
Spread * Growth, when spread > 0 -0.257 -0.219
(0.104) (0.103)
Spread * Growth, when spread < 0 -0.079 0.473
(0.791) (0.562)
Year Fixed Effects Yes Yes
Bank Fixed Effects Yes Yes
The Risk of Default
Are firms more likely to default because they
are in the priority sector?
According to the data from our bank, 2.5% of
loans in the priority sector become NPA every
year
For firms included in 1998, the rates of default
was and remained lower till 2002 (but climbs
sharply in 2001, after they get excluded from
the priority sector).
Cumulative fraction NPA
Size of the firm
Small Big
(1) (2)
1997 0 0.011
1998 0.026 0.011
1999 0.052 0.0229
2000 0.078 0.057
2001 0.118 0.0919
2002 0.125 0.137
Is public ownership
the problem?
Is public ownership the problem?
Cole (2004) exploits a natural experiment the nationalization of
banks in 1980 to answer this question.
The 1980 nationalization took place according to a strict policy
rule: all private banks whose deposits were above a certain
cutoff were nationalized.
He compares banks that were just above the 1980 cutoff to
those that were just below the 1980 cutoff, while controlling for
bank size in 1980.
The idea behind this comparison is that the relationship
between size and behavior should not change dramatically
around the cutoff, unless nationalization itself causes changes
in behavior.
Figure 1: Rural and Agricultural Credit
Credit to Agriculture Credit to Rural
.05
.05
0
0
Share of Credit
Share of Credit
-.05
-.05
-.1
-.1
-.15
-.15
-.2
12 13 14 15 16 17 12 13 14 15 16 17
Log Deposits in 1980 Log Deposits in 1980
Table 7: Deposit, Credit, and
Branch Growth
Log Real Growth of: Growth of:
Deposits Credit Branches Rural Branches
Post (1980-1990) -0.085 *** -0.078 *** -0.114 *** -0.181 ***
(0.014) (0.015) (0.017) (0.024)
Post*Nationalization -0.026 -0.012 -0.044 -0.066 **
(0.033) (0.036) (0.033) (0.031)
Nineties (1990-2000) -0.040 *** -0.027 -0.122 *** -0.219 ***
(0.014) (0.017) (0.018) (0.022)
Nineties * Nationalization -0.073 * -0.088 ** -0.053 -0.086 ***
(0.039) (0.041) (0.034) (0.028)
Table 9: Causal Effect of Nationalization on
Lending
Estimate of Discontinuity
1992 2000
Average loan size: -24.753 ** -143.867 **
(10.332) (69.784)
Share of bank's credit to:
Agriculture 0.082 *** 0.031
(0.030) (0.021)
Rural areas 0.073 *** 0.021
(0.027) (0.023)
Small scale industry 0.009 0.020
(0.017) (0.026)
Trade, transport and finance -0.073 * -0.037
(0.040) (0.031)
Government credit 0.020 *
(0.011)
Intere rate (residual) -0.007 -0.007
(0.008) (0.006)
Not much of a demonstrable gap
between public and private banks
Overall public lending grew at the same rate as
private lending in the 80s; lagged behind in 90s
No differences in lending patterns today.
Historically public banks favored agriculture/rural
borrowers, while neglecting trade/transportation.
No differences in lending to small-scale industry.
Closing thoughts
What is happening right now: some
speculation
Over the last couple of years, there may well be a shift.
Lots of innovation going on in the financial sector.
Big banks like ICICI Bank and SBI are becoming increasingly
aggressive
Private finance is growing fast
The stock market is booming.
Boom in retained earnings in the corporate sectors
Large corporates are borrowing heavily on the world markets
My prediction is that we will look back on this period and
conclude that the recent growth acceleration had to do with a
sharp inflow of capital into undercapitalized firms.
