'11—...—
| ‘ J "
“:4.“ u .- -.. -:- ... ~. «.. u a. .,.. .- ‘ Mn 4" '2' -I~- "a'

,.. uwun wuqvuan:'I--.~(€"§-““'\‘~"'EIEH «nun-4 em-n.......a........» ..~

 

   

L}

.I' ! 5“; ~‘ ,q I .\ w
. “I ‘1 »
,-,-' b7 ; i :3 "

 

This is to certify that the
dissertation entitled

PUBLIC INPUTS AND THE CREDIT MARKET

presented by

RAJALAXMI KAMATH

has been accepted towards fulfillment
of the requirements for the

PM) ECONOMICS

degree in
/ W - ”W

Major Professor’s Signature

M

Date

M 5U is an Affirmative Action/Equal Opportunity Institution

 

  
    

 
 

. LIBRARY
MlCht'gan State
University

 

 

PLACE IN RETURN Box to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/CI RC/DateDuep65-p. 15

 

 

 

PUBLIC INPUTS AND THE CREDIT MARKET

By

Rajalaxrni Kamath

A DISSERTATION
Submitted to
Michigan State University

in partial fulfillment of the requirements
For the degree of

DOCTOR OF PHILOSOPHY
Department of Economics

2004

ABSTRACT

PUBLIC INPUTS AND THE CREDIT MARKET
By

Raj a] axmi Kamath

C e u ' f‘ ntribute towards
Publi inputs like infrastructur SCd collectively by Irms also C0
r e frrn 'n c ‘|' ' 'm rtant im lications
red ° [he hete 0g neity among 1 S 1 an e onomy. MS has 1 PO P
ucmg

' etries
w ’ ‘ ' formation asymm
' ' arkets, hich are besneged b in
for allocations m the credit m y

' rs. This
h frms Who are the borrowers and the banks who are their lende
among suc 1

. c" “35
w ' ' ' ' dit mark
- . ,. - Uth and pnvate Investment Via the cre
cruc1al“rr11cro link bet een p

sea
been explored in the first chapter of this dissertation. In a model island economy 0: on
faring entrepreneurs, we trace the effect of an archetypal PUbliC 800d - a lighthous ,
the credit market equilibrium Of this hypothetical economy. Results indicate that the
effects of the lighthouse on the credit market equilibrium not only have an impact on the
Optimal level of public inputs in an economy, but they also say something about the
‘targeting' of public inputs in an asymmetrically informed world. Public goods that are
targeted to the low ability may dominate those available to all types. Thus, this C

hapter
contributes to the debate on the precise linkages between infrastructure and economic

develop ment-

T‘ he second chapter explores the role Of “social infrastructure” of an economy 0n
shaping i [5 business environment. Social infrastructure includes not only the physical
infrastruQ ture of an economy, but also its legal framework, business regulations, sc0pe

for corn“ ption and need fOr irregular payments by firms etc... It has been observed that

those economies ranking high On this social infrastructure index consistently attraCt
higher levels of domestic and foreign investment, as compared to economics plagued
with corruption and poor social infrastructure. In a simple theoretical model that
explicitly takes into account such factors, it is shown why a lender (presumably a bank)
would look to these economy Wide indicators instead of firm-specific indicators to
determine its lending decisions. It is concluded that in contrast to private signaling by
firms in the credit market, these factors will increasingly be looked upon as ‘PUbliC'
Sifinals, which improve allocational efficiency in the credit market.

In the third chapter of the dissertation, the hypothesis about public inputs and its

' . . 'de
effect on the credit market proposed In the two Chapters above 18 tested usmg world W1

. - ss
firm-level data based on a survey carried Out by the World Bank Group (World 3‘15““:

. de
Envrronment Survey (WBES), 2000). We Concentrate on two sets of constraints {ace

firms - financial constraints and the quality of public services, We Show that the quality

of infrastructure faced by firms crucially affects the financial constraints they face ill the

credit markets. Both Ordered Logit and Ordered Probit estimates Validate the

that . . - - . COHCIUSion
talgng care of all region spec1f1c and firm specific constraints firms f

a .
infrast ~ ' - Clng high
mctural constraints are most likely to have high financial constraints as

l- k - well' This
m ’5 S eon to be stronger in the case of smaller and medium sized firms.

F 0
r my Parents

iv

ACKNOWLEDGEMENTS

Thanks firstly to my parents, for bearing with patience - the peregrinations of my career.
Thanks to my Advisor, Professor Lan)’ Martin for having the confidence in this work — at
times, far exceeding my own. Thanks to my siblings, for providing the succor when most
needed. Thanks to friends who were there to listen.
[began this work with the spirit, “De omnibus, es dubitandum” (Of all, one must doubt)

T “101’" I have succeeded-

EV

TABLE OF CONTENTS

List of Tables ................................................................................. vii
LiSt 0f Figures ................................................................................. vul
Chapter 1 Public Inputs and the credit market ....................................... l
l .1 Introduction .................................................................. 15
1.2 Model .................................. . ......................................
1.3 The Capital Market and the Separating Equilibrium .................. 10
1.3.1 Public Inputs and the Separating Equilibrium ........................... 15
1.4 Switch to a Pooling Equilibrium ........................................... l9
1-5 Switch Point and Targeting 0f Pilblic Inputs. ........................... 23
1-5- 1 Public Inputs and the Pooling Equilibrium ............................... 27
1.6 optimality Rule for Lighthouses ........................................... 29
1- 7 Conclusion ................................................................... :1
1 -A Appendix ...................................................................... 44
LB Notes to Calculations. . - .....................................................
Chapter 2 Public versus Privalte Signals in the Credit Market ..................... A25
2- 1 Introduction .................................................................. 9:3
2.2. Model ............... .. . .. .................................................... 5\
2. 3 Capital Market Equlllbnum ................................................ ,5
2-3 - 1 Equilibrium when the banks can distinguish types. . , . . . . . . .. . . - , , 5
2-3.2 Asymmetriclnforrnation equilibrium................. ,_ .--. 56
2-4 Sociallnfrastructure and Credit Rationing............, 58
2-4.l Eifectofadecreaseintheparameter'y........... -""””"“"'--
2-5 Conclusion ................................................... W~-.6]
2-A Appendix. . - ' ‘ 62
Chapter 3 Infrastructure and Financial Constraints — What do ‘ ‘ 64
Firms have to say? ................................................
3, ‘1 Introduction ...... "‘67
3 _ :2 Data and Methodology ................................................... . . . 67
3 - 3 Regression Estimates ................................................... . _ 7O
3 - <4 The Infrastructure —— Finance Link ...................................... 72
3 ~ 5 Estimation over sub-samples — Characteristics versus """"""" 81
Coefficients ..........................................................
3-6 ConcluSlon ........ 88
3 ‘ ‘A Appendix ..................................................................... g”:
Blbllogt‘a hy ................................................................................. 97

vi

Table 1

Table 2

Table 2.1

Table 3

T able 3.1
Table 4
Table 5

Table 6

Table 7

Table 8

Table 9
Table l 0

Table 1 1

LIST OF TABLES

. ' ’ d Infrastructure 2
i n Fmanmal conStTalntS an ............ 7
correlatlon betwee ......................................
Constraints ....................... 73
Ordered Probit Estimates ..............................
' . traints facing firms 4
i . fFlnanCla] cons ............... 7
sample StatlStlcs o ...................................
(Probit Model) ........... 75
Ordered Logit Estimates .................................

1 Statistics ofFinancial ConStl‘aints facing firms ...... 76
am e ...............................................
(SLngt Model) ..........

11 Probability of haVing different levels of Financial .......... 79
goiritaraints (Probit Model) ........................

11 Probability of having different levels 0f Finam‘al , . . . , 19
era ..................................
ggnstraints (Lo git MOdel) .........

ffect of InfraStructure on the probability rallkings 8‘
gi‘eFemancial constraints (Probit Model)“ . . . . . . . . . . - .

ffect of Infrastructure on the probability rankings
6 . . .............
TOhfeFinancial constramts (Log1t Model)

Q.....

‘ ~ . .82
Finance as a constraint for firms by difficulties faced
As regards Infrastructure .............................................. . ~ . .84
Ordered Logit estimates over sub-sample (infr_d = l) ............ 87
Ordered Logit estimates over sub-sample (infr_d = O) ............... 88
Predicted Probabilities ...................................................... 90

Vll

Figure 1
Figure 2
Fl'gure 3
Figure 4

Figare 5

LIST OF FIGURES 14
' t .........................
.librium in the cred” Make ts ........ 22
ting Eqm ' Contrac
Separa t dominating the Separating ..... 24
. comrac ................
A Poonng .............
. ........................ 25
Targeted PUbhc Inpm .......................
............................... 60
Pure PUblic InPUt ...... . . ..........................
t re and Credit Rationing ......
Infi'astruc u
viii

‘

Chapter 1
Public Inputs and the Credit

Market

1 . 1 Introduction
. . 1e of public goods which increase productive
d mterest 1n the ro
There Seems to be a renewe

ternativel kno ' .
capmities of private firms [71]. Such goods, a1 y W11 as znter'meQVZate pub/2'5

tion decisions f ' .
~ - ts aflcct lrlvestln 00’
900618 or Public znpu

B ad examples include public infrastructure like roads 3-11
a COllthive manner. ro

. - untries and rovisio -
' k 1 1th care and educatlon 1“ many CO ’ p n or "Hem,
servlceS 1i e lea -

h h . t'tutes or agricultural extension centers. The macroeCOHOmic r01

t rough researc ms 1 e
f 1. t Capital has been dealt extensively usmg aggregate econometric
0 Sue}: pub 1c sec or

S l alysis however fails to clarify the precise “micro” linkages bEtWeen
model 1], nor an‘
& [5

b1' ' puts and the nature of the production PTOCCSS- Much of the analy-
provigi (311 of pu 1C 111 ..

S S I“ SO i, based 011 t e 353L111 P50” 0f DBIf t’ .IllOI ' amon
. i ‘ a] lIIlpllCl ly h ‘ l 1 8C 1 mathIl g
l t. 11is area IS

1

participants. It is now amply clear that public interventions in an imperfectly infonned
world have qualitatively different implications compared to interventions in a first-best,
perfect information worldi44l- This paper therefore, plans to pursue one “micro” link
Where the role of the public input is analyzed in the absence 0f perfectly informed mar-
kets. It traces the effect of public inputs on private investment decisions of heterogenous
entrepreneurs who have to borrow in credit markets characterized by asymmetric infor—
mation. In turn, this paper also addresses a key question in public sector economics
today [1?] _ In a world where asymmetric information is endemic, “What type of public
pOIiCies can relax the self-selection COHStraints?”

. . we

Our model economy is a small: Island economy of sea-farmg entrepreneurs and

. x, eddy
trace the effect of an archetypal Puhllc good — a lighthouse, on the credit “,3ka
. . . 090ml
l1br1urr1 Of this hypothetical economy. An entrepreneurlal project in this island 60

involves entrepreneurs undertaking a sea-voyage. Entrepreneurs differ in their knoWledge

0f the hazards err—route the sea-voyage ' efficient entrepreneur
S having better k
nowledge

about, the hazards compared to the inefficient. POtential entreprene 11

funds from the capital market in order to undertake the project Th 1 O borrow
. e GHQ
er

credit market cannot distinguish, ex-ante, between the efficient and the in 8 1'1] the
8

Cjfin

trePl‘etneurs. A lighthouse reduces the risks associated with such a Voyage a ear
1)

di

I‘ . 110

the D Qbability of undertaking successful entrepreneurial Projects, but it d teases
068 S .

asym I'rnetric way, It points out potential hazards at sea to the entrepreneurs
Undertak-
lug

the DbQJ-ect. Since entrepreneurs differ in their abilities, the lighthouse ben fit h

fiClell t entrepreneur more than the eifiCient. The lighthouse thus

in ~ ,
”83be the degree of

hOmQ geneity in the abilities of entrepreneurs, This key result affects the credit mark t
e

2

 

equilibrium in this economY-

The intuition behind this pal”?r is gleaned from field studies carried out in the rural
Credit markets of several developing economies[47]. Information asymmetries between
borrowers and institutional lenders like banks tend to be glaring in Shhh markets as
formal credit institutions are still in their nascent stages. One econometric study, for

examplelnl, covering eighty five districts in thirteen Indian states has shown that there

exist crucial linkages between public infrastructure and the process of financial interme-

djetion- Government investment in infrastructure like roads, irrigation and reglflated

markets which reduced the 1' iSkS that faliners faced, also facilitated the wipe-“51°“ 0f

. . . . t in
commercial banks in the rural areas. This result implies that government mvestrnen

. . . etween
infrastructure, by reducing produCtlon risks affect the pattern of information b

. in
. o o {‘86
the lenders and the borrowerS~ Pllblic inpUtS, by reducmg InfOI‘Ination asymmeh

the BCQnomy can thus play an important role in improVing the equilibrium in the Cledlt

marke t.

The credit market model we consider is characterized by inform t'
a 1011 3S

which are ex—ante. We essentially build on the model 0f Bester[16] Whe
, re call

a cat, 8.1 ytic role as a private signalling device in the credit market eQuih‘brium

game Igeing played in the credit market on the lines suggested by Hellwr'g[46]
- D

u, . . -
P011 1:110 degree of heterogeneity 1n this economy, the effect of the lighth
‘ ‘ Ollse

on both
separating and pooling equilibria in the credit market, will be considered T ,

ell - ‘ ’
ectx of a lighthouse are Summarized 111 Samuelson’s EMRS, while the ind‘ t if
irec e ectS
St . . I . .
em fr cm the asymmetric Information in the credit market. By seDal‘etely considerin
g

these indirect capital market effeCtS 0f the lighthouse, we can also Say something about
the ‘type’ of public goods that ShOUId be pI’OVided in an asymmetric information world.
As our intuition suggests, the results of this paper consistently point out the increased
benefits from a. lighthouse via the credit market. We see that public inputs, by lowering
the agency costs which define a separating equilibrium in an imperfectly informed market,
Shif t the second best Pareto frontier towards a more desirable pooling equilibrium. While
they may not eliminate information asymmetries totally, they certainly expand the size
of the information set available to the economy. This link between public and private

aneSt’nent is crucial, since finanCial mtermediation is increasingly playing an irnporl’a‘nt

' - - s . t 0‘
r016 In private investment decisionb 0f most economies[40]- ThlS additional benefi
‘ oliCies
a PUbllc: input in a second beSt economy should 1101? only have a bearing on
X)“blic

determining the optimal level of public inputs, but 3-180 0D the ‘targeting’ of suC

inputs -

The paper is organized as fOllows - Section 2 introduces the mod 1 I
e - 11 Section 3 we
discuSS the separating equilibrium in the credit mar ket. The effect f b
' ‘ 0 pm 11

the Separating equilibrium is given in SUbSGCthD 3.1. The iSSue Of th Inputs 0”
9 pool.
I

rlurn is dealt in fair detail in sections 4 and 5' It is seen that increas'
1mg

Publ iQ goods of particular types affect existence of the separating Contra t Vision of
C S -
. 1
it . -
31 111 arket- Thls gives us some results on the targeting of Particular t G can
y es
of Dub

in . f‘ '
I)” t’N . Section 6 enumerates the ban fits of a lighthouse on the economy 1
1Ilder bo h
' t th
I 0

re ‘ . .
glrrl Q5, separating and pooling. The necessary modifications to the Samm31 R
son 1118 for

the . . .
Db .ov1310n of public goods are suggesmd The paper ends With sec tion 7 h
, w ere We

oi . .
D Rt. (31% areas of further research and give the COnClllSlOIlS and some D011 cy implicat'
10113

4

of our analysis.

1.2 Model

' ' ' ‘ 3 -- ct to their
Agents. WC corlsider a small island economy in Which agents differ With respc

innate ability a A proportion 7 of these agents is of low ability, denoted by 0L and
(1 ‘ 7) is of high ability, denoted by 011- Whether an agent is type aL or a” is private

0 o o O c ' 1 e
infonnat ' Agents can take up entrepreneurial pro J ects in this economy which an0 v
1011.

‘70 te WhiCh
' a. ya. 8 Y the hazards en’rou ,
Sea V033_ge_ [he IISkS on SUCh g are defined by

‘ i t, agents
are N in ber So ability in thls ecoIlomy corresponds to the knowledge the
num . ,

. the“
' ‘ er 1“
haVG abQ t the hazards on this Sea’VOYage, and the potential entrepreneurs dlfi
1.1

a 0“
. we
knowledge about the hazards. These N hazards can be thought of as being. or

t6 ”A

a scale of visibility- The most VISIble and obvrous azar s are known to all a

the less, obvious are known only to the efficient types. Agents of type a H are aware 0f

M H hazards and agents of type 0L are aware Of ML hazards en route. ML V M /V
fl < .

Additionally, we assume (ML) C (MH) The type a” entrepreneurs are aWab

hazards known to the type 0L.

Q
012-111 the

A 11 agents in this economy are identically endowed with one unit of labo

1" (I)

o ‘ O Wh -
they ‘ ' . - We assume initially that all agents are also ident' 1C}:
S hpply inelastically. many

with

endOWQd
With a physical endowmentw. We RSSUIIIG that w < l, and u) can be Used e

ither for
consLl 1" leti on or investment. These agents are risk neutral and live for two Derio

ds. They
prod 11 C6 in the first period and consume in the second period. Agents

have two distinct
C110ng 3, to be workers and work in a ‘TOutinc’ activity, or to be potential

entrepreneurs

Cf!

 

d bark on an entrepreneurial project involving a sea voyage- M’Orkers are assumed
an em

1y only and for their labor, each worker gets a deterministic return 2,. Efficient
SUPP

k s get a higher return denOted by ZH, as Compared to the inefficient agents who
wor er

get Z L. We assume,
A 1 Z” > ZL.

This assumption implies that efficient types dominate wrth respect to this ‘routine’ ac-

tivity also and as we shall see below, this assumption

is '
Important to the results in this
model.

' their labor with one
The potent“1 9“ rel" It of capital (k) and
° ' t. One unit of capital (1:) is .
the entrepreneurial ProJeC .
embark on . Vested in the prOJeC t
one period ahead of time, and it fully deprec1ates after one period

- f - . fp r 0duetion. Poten-
tial entrepreneurs have to borrow this one unit 0 capital (1:) In 0

er to
undertake the
project In addition, we assume that they can only use In as “mater

8.1 to O .
' ' hi wreck, the f . Q11) ca “8,
('1’) for investment. If the prOJeCt ends in a S p- Y Orfelt t eir €011 P 1
BanS. Banks are special firms WhiCh oper ate as delegated monitor

tefa].
S for '3.
' ' ' k ny’ende
as deSQ ibed by Diamond [35], Banks In this economy are I‘lS —neutral and act rs
r

. . Betrand
Com D Qt' tors in a market where they obtain elastically supplied funds. We nor
1

gross ‘1 osit rate at which the banks obtain these funds to be one. Since
813

We 388qu
. fits in e uilibrium .
bank§ to be Betrand competitors, they make zero PTO Q on the prolects

they I
13

find.

' ts" The returns from the entrepreneurial PFOlCCtS have the f0110Wing character-
\fijcc ‘ '

$ng \ they are:

(a) Uncertain - A successful PFOJ ect yields y > 0, while an unsuccfissful project, which

involves a ship-wreck, yields nothing.

(b) The returns depend Upon the ability of the entrepreneur. Agents of type an (ell
ficient) have a higher probability Of making the project a SUCCESS, as compared to the

type a1, (inefficient) agents. We make a Simplifying assumption that a type an agent

. A!
has a success probability given by ’Nfl’ and a type aL agent has a Success probabil' t
1 Y

iven b Mi. Thus the abilit of the entrepreneurs in makin a .
g y N y g SnCCeSS of their PTOject

is linked to their knowledge of the hazards on the sea journey, ie, M . '
’ 7w 2 6 {H,L}. The
efficient agents get higher expected returns from the project as

. Compared to the ineffi-
cient agents, A—fiiy > iffy. In this model, followmg DeMeza an

, _ . _ Webb [28], we assume
that the distribution of returns to the hlgh ablhty (low risk) bOrr

. - - er GXhi its first order
stochastic dominance (FOSD) over the distribution of retums t b
0

_ the 1 .. .
risk) borrower} Here, the high ability types get hlgher return Ow ability (high

in an -
undertake. The Opportunity cost of entrepreneurship are higher f aetlvities they

. . . ”10
SO: an efficient type will invest "1 the PFOJeCt only If the net returns f re efiicjent
r011) '

Exceed his opportunity costs.2 e project
(C) h igh yielding - We explicitly assume ‘high yielding’ to be,
A 2 {H11 y — 1 > 21. The expected value of the net returns from the proje
JV Ct exceed th
6

0
Eva ktu'n’ity cost of the project to all agents.

 

1b ' ' '
{\is is in contrast to the Stiglitz-WciSS [65] model» Where d‘Str‘bumn Of mums ‘0 the high risk is

a m . .
2c 3‘ {E preserving spread of the distribution of returns to the low risk (SOSD).
b is assumption is also in consonanCC With Spencc’s labor-market Signalling model, where the

reset ,_ . . . '
\ ;:\_ tion wages of the high ability are higher.

While 91%; - 1 are the expected 1' etUrns and 1 is the value of the l1:)01'rowed capital, this
assumption implies that there exiSt incentives for all agents to undertake the project.
Public goods ‘9’ in the nature of zinfnwit"‘ucture: The economy has to be provided
with ‘g’, which has the characteristic Of a pure public good, in this case ' a lighthouse.
It is non—rival and non—exclusive in 1158' HOW g enters into the dECision making process
will be explained below. It is assumed that the level of infrastructure cgv plays a Critical

role in determining the success probabilities of the project. A lighthouse points out

additional hazards on the sea—route to all entrepreneurs. Since the hazards can be ranked

in increasing order of their visibility - a lighthouse win typiCal

y result only in two
possibilities,

(Athe targeted public good: a lighthouse which will brill

. . '30 light one hazard
known to the efficient type, but not to the IneffiClent type, 0,

(B)The pure public good: a lighthouse which will bring to light on
e h

azard
before to both types.

3 unknown
In Scenario (A), We are restricting ourselves to public. gOOdS WhiCh directly

. . Th 1' hthouse here directl Qnefit 0
the 10w_ability types in the economy e 1g ’ y benefits 0 my

.3? the
° t
a5, Si 11 06 the type a H was aware of this hazard before. J’De
“21% define the effect of this lighthouse (A) as per the followrng, Vi e {11,14},
dU,‘ __ BUL ___L.dM where —LdM = 1. a d
dam) ~ aMb dQM) ‘19“) nd fl“

dgp‘) : 0
SMILES“ ‘ (B) On the other hand is him general case ' public gOOdS benefit both Woes
1‘10 1 ‘ ’ ‘ ’ " ‘ ‘ "

in tl). §\ onomy efficient as well as the inefficient. With the provision of the lighthouse,
‘ eC ’ '
«3" 4 and A! L increase by one in this case. We can define the effect of this lighthouse
H

KB\ 3kg per the following, Vi E {H, Li’

whef‘e ail/L zeta—:1.

_ (9U 4M an
(ill ’1‘ \L __L_ -l I (“Wu
49(3) (19(8)

_——L—
(19(3) BAIL (19(3) 8M” (19(3)

We then have the following,
P sition 1 In both scenarios, (A) and (B), the lighthouse increases the
ropo
h eneity in the economy ' However, this increase in the homogeneity is
omog

3
higher with 9(A) as compared to 9(3)-

The intuition behind this DTOPOSition is that the lighthouse affects the productivity of
- '1 th (1‘
entrepreneurs in an asymmetrlC way. 80’ Whl e e fleet effect 0f the lighthouse on

, . . . is obvious, we haVe to consid - . ,
the PTOduethlty 0f entrepreneurb er the ‘indirect effects

eneity in the economy. We tr
which stem from reduced heterog QQe this indirect effect
- - ’derin both
. t. Additionally, by conSI g t .
through the Capltal mar ke h eSe Effects, the direct
- ble to compare 1: e two seen ,
- ' ct, we Will also be a 1 .
and the indire OS (A) and (B) This

will basically tell us something about the «type of public goodS at:
ShOUId be built -

t

 

\ . f homogeneity, then pr . . a1],
Mi. 3 the degree 0 ODOS
3” We assume M" to summarlz "no" (1) Stat
$3 that A!‘

. ' _ this ro erty. To take a con
‘nereasing Not all Pnblle lnPUtS may have p p Crete example . My
. ’ In

 

p "blic ' es like setting up Public research instituteS, 01‘ DFOVlding information about
SeerC

is}: -
_ _ J’Iejd-
agricu 1t. 1 techniques and inputs to farmers (both actively pursued by the Indian gOVernme mg
um I" dun
he

head] on benefitin th
the ‘EF . , . 1970’s) the focus could be SDBCI y g e most
Q‘en revolution in the ’ produCtiVe

making Our

en: in the economy will fall. However, 1)
seetorg In this case, 11%;, the degree 0f homogen y y

\

' - v ture the nature of benefits derived fr .
88511”) M - .~ ‘5 fit any hope to cap Om "‘0‘
. —J- ~ asma - we es 9. 1 st
Q tion Of My mere 0

bastc i \ fr aStructure goods and services.

1.3 The capital market and the separating equilib—
rium

In this section we consider the Capital market Where the information asymmetry is ex.

th ' - ° ' M M
ante viz., the borrowers know whether 911" success probablhty ‘3 TV“ or —N11, but banks

only know the average ability of the entrepreneurs applying for 108.113
The optimal lending contract: Assume there exists only a Standard debt COHtract,

for investing funds (issuing equity is costly). As in the model of BeSter [16] banks offer

a loan comract consisting of a pair (r,c) where 7' ‘S the gross int

1‘ est rate charged and
c is the collateral that the borrowers are wrlhng to put Up- Simil

. . . r to the Besanko and
Thakor model [15], we make the followmg additional assumption

A 3 Collateralz'zing is costly. Collateral of value Ci t0 the 5077b

er .
92068 t t b
value of fig, in the event the borrower defaults, where fl 6 (0, 1). 0 he ank a

Note that apart from having real world justificatlon, thls asSunlption is 1
a SQ

' ' C‘r .
the existence of equilibrium in a rlSk neutral enVlronment[26]. As in med Hera] for
Q1

' ' ‘ - git»
[46], the a 3 stage sequentlal game IS bemg played 11} the cre di t "161 en by

”re:
gam . d ‘ actions of borrowers It ' . This
e prlicitly takes into account ynamlc re ‘ - 13 Played as f 11
St . _ , . ,
age \1 : Loan contract offer made by bankS, the uninformed player in the game, (Ti, Ci)
$093 11- The informed players, the borrowers: Choose to apply for contracts they View
as m '\\:;t attractive. Each borrower can apply only for a single contract.
Stags: 111° The Bank may accept or reject the loan applications they have received in
8&ng

II.

10

The Optimal contract (fa-fir); Vi 5": H1 L is a set of contract offers which determine the
equilibrium in this 3 stage game- i.e. given (73,61) Vi = H , L

- Banks make zero profits on each contract and

- No bank has the incentive to offer a different loan contraCt than the ones offered.

Since this game is sequential, the equilibrium considered Will be the sub—game perfect

or sequential equilibrium. The usual COHdition that each agent’s Strategy be the best
response to the other agents’ strategies is applied not only to the overall game, but
to every decision node in the game tree, regardless 0f Whether this node is reached in
equilibrium. We define Uij to be the utility to agent of type 2' 21pm)?i

11g for a loan contract

meant for type j, and it is given 38:

Mi 1%
Ur]. N (y—m—u— WM —2;

 

ll

The problem for the bank is,

Max 7ULL + (1’7)UHH sat.

{rs :Ci}

. Mi . Mi

(1) the? Zero profit constraint for the banks: W“ + (1 — 7v— BCi = 1.
1.. _ .
(1) t1” 1% self selection constraints Uii 2 U 0'-

B Q fore analYZing the asymmetric information case, we solve for the perfect info
rrn

‘

thn . - _ .
Q ‘ -r g a benchmark. The self selection constraints do not bind this
Q58 3* case. We
In ' - 3 . -
snark-1719 U m SUblect t0 the zero profit constraint of the banks (1), and the assumption

that “Cir 4 1_ We see that [See Notes to Calculations],

Ci = 0 (N0 collateral).

11

\l
\

Ti

1112‘ i : {H,L}_

In the perfect information case, the interest rates charged by the banks reflect perfectly
the risks (inverse of the entrepreneurial 8Lbilities) associated with the project. The low
ability entrepreneur is charged a higher interest rate (TL) than the high ability type
(r11). These contracts cannot be optimal in the asymmetric information case, since bOth

types would prefer contracts meant for the efficient types.

To see the equilibrium in the Asymmetric information case, totally differentiate U

to get the slope of the indifference curves of the borrowers in th0

r‘c space,
d7} N — Mi

21—- Mi
6‘ (1.1)

(1) establishes both the negative slope and the single (3033ng pro

ebt .
- - . y of the indifference
curves in the r-C space. The indifference curve Of the memment b0
1” 1‘0 .
- Th efficient b Wer IS steeper (has
lOWer MRS) than the effluent borrower. e orrower

S 1e .
on account of a ship-wreck and lose hlS collatera - 0, Of a given dec 0 efault

reHSe in
rate, an efficient type would be willing to DOSt more collateral then the in ejnterest
e

. . Qic-
eXPIai 115 the reason for the higher MRS. Likerse, the expected returns to th at
Q

. bank
18 $th by 9,11%“ + (Ljvh’ll)fi0l Totally differentiating it we get the slope of the . (Rb)

 

SOI‘evenu
€11er . - ace e
for the bank in the r c Sp ,
dri _ —L3(N - Mi)
222: _ M1. (1.2)

3.. \istenee Of a separating equilibrium requires:
(a) b :t e isorevenue curves of the banks should be flatter than the indifference curves of
‘1

thek) Qrmwgm In this Case, assuming 13 < 1 assures us this.

12

(b) A condition on the composmion 0f the economy, where we ass ‘Jme 7 is high. This
assumption - the presence of large number 0f 10w ability individ 118.18, ensures us the
existence of a separating equilibrium (See Rothschild and Stigtitz [60])

For deriving the equilibrium contr act (721,61), we see that 11) does not impose a. binding
constraint on collateral. For the equilibrium, the only self selection constraint which
is binding is U LL _>_ U LH — the constraint which applies to the low‘ability entreprenem
mimicking the high ability types.4 This is used in Solving

f .
or the Optlmal contracts and
we get, [Proof in the Appendix],

Proposition 2

" N
. C :: (I
c = 0 H c \ M
A _ N T‘ : ~N ‘MHD
TL " ML H M” +86

6' (1 \ A)
. . My '
In the process of self-selection, the low-ability types get the same (:0
“tree
, . . . 1:
under the perfect information case, while the hlgh-abllity have t o dist- 38 they would
111g Dish

by willing to pay COllateral. Collateral in this economy has a (3031;, Sine themSBIVes
borrowers pay higher interest rate than the high-ability borrowers, they hav
e

- 1) in
to m‘ 1111c as the high-ability typeS- They must therefore be deterred from the c Centive

C . . . O I C Q g t
Om“ act meant for the high ablllty typeS- T1118 ‘5 aCh‘eved by requmng the hi be

t gh‘abuity
o . _. .
DQ§ t Collateral, since it IS onerous for the low—ability to post collateral.

 

 

4F . '-
}\ 6 Optimal solution is obtained by COIlJeCtllrlllg at first that UHH > U H L, so that there exists
only _ _ _ _ . , . .‘ . b, = _
Q ‘ixue self—selection constraint which 15 binding, 311d Wh’Ch ‘5 given 3 ULL ULH. After SOlV'lng.

[or “X“ op‘imiyai ion problem we can show that the solution satisfies our conjectureisee Besanko and
- 7 , . . , .

T\\B.\§Q b [.15] )

13

F' e 1 Separating Equilibrium in the Credit
igur -

  
 
    

 

 

Market
r (FL, cL=o)
‘1
(TH, CH)
TH BL
Bu
C

' ‘ shown in the
Collateral sorts borrowers by type. The same lb figbq‘

I
f t th : 0V8 Where in the
ac (1 re ers o - ,
interest rate (r) and collateral (0) SP e, the H e "Rh

eren Ce c
are flatter than the indifference c
high ability entrepreneurs. They

. W
‘ . , 1‘ profit 111108 0
BH refers to the 70
t reneurs denot d by L,

0w
‘ 'brium contracts for the low-e .
boerWerS (TL) CL) and (771,611) are the equill fliclency and
the I‘ ‘ . es CCthEly-
l 1 gh—effimency types r p

abiIi ty

q 'v ' the credit market, the equilibr‘m
se ' ~ ' t1 e contracts in 1
If i. election of the respec
the

l l ’ ( t1( l‘ K‘ ' “(:11 (in; l)(,l()Uh‘ )
% 01 th(’ by" t p n‘ of ’n ' 'I) ,n(’lll , ar‘; g1
.~ 5, ’ I]

A ML ’1
ULL = _N—y

N - Mu)
‘ 5313/ -1-afli<1—m<(—-—N—~—)1

5
I
ll

14

 

We see that the low ability entrepreneur ends up with a payoff eq 11a] to what, he would
have got in the perfeCt informatI'On case. On the other hand the high ability 611-
trepreneur’s 1333“)“ is lower to the extent 0f the signalling costs which he

has to incur in

the form of collateral.

1.3.1 Public inputs and the separating equilibrium

Lighthouse (A) -' The targeted Public inputs We begin initially consid
. - c6
lighthOUSe (A), which is directly beneficial only to the low ability types“ Sm

are

. e
Increases M1,, the hazards known to the low-ability types. The total befi

calculated as follows acted

. 6 .
(1) “Direct” benefits: All the low-ability entrepreneurs face an inCrease in ”b "13 T‘ms
0“ o
. . . . 0‘531
returns due to alowermg of risks In the sea-voyage, given to be 116 whiCh 15 P

equals the marginal benefits of the lighthouse in a perfect information world as well.

(ii) “Indirect” benefits stemming from the capital market These indirect benéfits are b
. a-

SiCally of two types: (a) The benefits accruing from lower interest rates due t

 

 

10 . ‘ _ . . maIgl'nally
Wer I'lSk of a ship-wreck. The hlgh-ablhty entrepreneurs do not get the b
QIzefi
t of
duCCd interest rates, but they benefit from (b) lower Collateral costs wh~ re-
’ 1o .
as, 18 given
aUHH 86H [( N — 1W
1. ~— 1 —— [3) (‘5.
39M) aM” N )1 (1 3)

We See that 3%}; < 0.[See Notes to Calculations (HH- Therefore, we get BU

N111
. . d9”) > 0' Thlls
by \Oworing the amount of collateral that, 18 required of the efficient tVDog

3 public #100
(ls

' 0 fi ,3

also reduce the “Slgnallmg costs .

15

This benefit of the lighthouse is peculiar to an asymmetric info 1717.52an environment,
where the outcomes are pareto i“Efficient- The direct benefit of the lighthouse to the
high-ability type is zerO, because he has previous knowledge of the additional hazard.
However in the credit market Where the banks are unaware of types: he has to p03t
collateral in order to signal his type- These Signalling costs are directly related to the
degree of heterogeneity in ability. The higher the difference in the abilities of the 'm:
. 1'
types (2%,: is low), the higher is the cost borne by the high—ability entrepreneur m w
Of COIIateral. A public good which increases the ability of the ineficlen
degree of heterogeneity in the economy. The intuition is that by decreas'mg‘ 6 €06

. . I) . ‘0
heterogeneity, the lighthouse relaxes the self-selection constralnt (ULL 7 U file ‘1

o . . a .

seli-selection constraint is relaxed, mimicking is no longer seen to be that 6&8 “A l“
. ‘ . f

the inefficient, and the costs of Signalling to the EffiCient is r e duce d. T1115 15

the lower collateral for the efficient types.

(iii) “Entry effects”: These arise in the case of the inefficient, entrepreneurs, d‘ue to lower
interest rates.
(ii) and (iii) are benefits of the lighthOflSe WhiCh accrue due to the inefficieh
c - . . . 01¢?st the
r ed1 1: market, arising from asymmetrlc Informatlon- Both these additional

. . U
have to be included in the Cost—Benefit analysrs of the lighthouse, and they fir St

di
‘ SCUSSed
“\ SeCtion 6.

Lighthouse (B) - The pure public inputs Here, the lightl1011sc benefitg

b
‘ . . 0th tyDos
of entrepreneurs. The benefits of this lighthouse are given by,

(t) “Direct” benefits to both types: Direct effects on the productivity of b th t
O ypes are

given by the increase in the expected returns from the projects- In the Gas f tl l
e 0 1e CW.

16

ability, the eXpected returns rise by 1, which . .
N IS the Same as In the Case oft/16 [get/101158

A . For the hi h-ability, his dire (3t product' '
( ) g lvlt’y benefits (Which were 2310111 the previous

case (A)), go up by the term, (716‘ + Qfifilcfl).
(1i) “Indirect b€n€fit5”: The indiTeCt benefits to the 10W ability accrue from lower interest
ability, the

rates, whiCh remains the Same as in the Case of lighthouse (A), For the high-

total benefits have to include the “indirect” benefits when stem from the capital market.

The tOtal benefits of the lighthouse (B) to the high-ability is given by,

 

 

 

aUHH __ g (1-5) _ 50H 5611 N—MH

The total benefits critically depend on the indirect ‘ capital’ market eff t given by the
CC 1 .

t8 __ acfl Beg _ 91.21.5111 ggi °
rm, {tam + aMHHU [3)( N )l}- AS we saw’ BML IS negative, but on the other

hand 2211. is positive [See Notes (11)] Therefore this i -
3 6M" ) ' , ndlrect . t eficct
capital 1113.th
being me - - - 'lld d the rel t' -
gatlve or posrtive w1 epen upon a Ive weights We
of these two terms"

then have the following proposition [Proof in the appendix],

Proposition 3 Indirect capital "“1"th benefits ”flig’lthOUSC (B) l
, va (ted 6’
qua/[y

by all types, though positive, are of a 16888?" magnitude than the ca 't 1
pz 0

, . market
benefits of lighthouse (A), valued only by the mefl‘iciem.

1126 ~ . °
1 ‘7 i 1 is th‘ I publlc input which has a greater impact on increasin
It u 8 the homo.

gene ‘ . - . .
1 ‘5’), in the economy leaves the ineffiment Wlth a lower incentive to mas
quel-
ade 1
' n the

0886} . .
sf lighthouse (A), the increase in homogeneity is higher compared to 1-
lghthOQSG (B)

Th . . .
el‘gfore with the prowsron of lighthouse (B), the resulting slack in th
e self-

selection

eon ‘ . -
St. 1‘ aint IS 1955, and the capital market benefits are lower in magnitude
as compared
to

17

the lighthouse (A).5 One has to the“ compare this capital market externa/ity W171] the

direct prOductivity benefits to find 011t whether the total benefits of the lighthouse (B)

exceed that of lighthouse (A).
(iii) The entry eflects; For the low ability, the entry effects rem

ain the same as in the

case of lighthouse (A).

 

 

 

 

     
    
 
  

 

 

 

   

SEPARATING EQUILIBRIUM.
Lighthouse(A) Lighthouse(B)
‘Direct’ effect (+) ‘Direct’ effect (+) ‘l
31, Interest rate effect (+) Interest rate effect (+)
\
a“ Capita\(:/1+'¢1)rket effect 3:32;; 13:31"; (e 2e“
4,

  

 

 

The fiEUIe above summarises these conclusions (++ Indicate benefits of a gr t
ea er mag-

nitude as Compared to +). Under a 88133r ating equilibrium, the low ability e11
trepreneurs
) and (B). The productivity and

are i I1 different thWeen the lighthouses (A .
the IntCTCst

rate I enefits they get in either case are the same.

5
F 5 is reason essentially sums up why a
credit market. There is no relaxing of the self-

public input which is more valuable to the

haVe ' '

.o - s ill-overs m the
_ wit} ML decreases. So any public input which increases the abiht 0
1 Such a good, M“ y f the efficient mere

than t 1‘ t of the inefficient increases the degree 0f heterogeneity in the economy. AS mimicking is
a made

active the collateral costs in the credit market increase.
1‘ a

Incite ‘tt
l8

- 1y, on
d epends assent/a1
between the types (A) and (B) then
' ' to chOOSe
The decrsion

~ . . . ta] mar
th

ductivity
. . 'rect pI'O

. th derive pOSlthe d1 .
with lighthouse (A), ey edlt
I. .

' ' he
the efficient types IS lug

- in the cr
' uilibrlllIn
(B) Suppose there exists a separating 6C1
'th lighthouse -
benefits W1

' 'nput
ted public 1
- the targe

has funds to build a lighthouse,
ent
d the governm
market, an

.v. at a
on) [)3 e . . thou (
d t the [Nile DUbliC Input (llgh 59
i
. A)) may
(lighthouse (

. . ‘ on ould

' ' 5.
‘ ed in section
t benefits, which 18 explor
capital marke

' equilibrium
“ itCh” to a pOOllng
SW roge’
1.4 The SS hem
in lighthouses, the entrepreneurs are made 19
‘ ting
hat by Hives
We have seen t

ver
Howe
.n ter 8 of collateral are reduced.
(1 the sorting costs 1
I I n
(is ability, at
“0118 as regar

, - in
.t can jeopardize the existence Of a Separating equihbrlumthe
this increased homogenel y of a separating equilibrium depends essentially, on
the economYlfiol’ since existence tiate themselves from the inefficient by Signalling. We
effici ent types wanting to dlfleren the public inputs and the nature 0f the equilibrium.
. en
now turn to the relationshlp betWe ilibrium denoted above, equilibrium pabhofi's to the
. equ
WQ saw that in the Separatmg

entrgh reneurs are.

A iii—Ly ’1

ULL = N (N - NIH)”
‘ M”y , 1 —éH[(1 —fi)(—N~
”W = —N—

D Q _ _.___,—— a w . the ‘sortin , cost, hi

fi)((N—M"))l s S, here S refers to g S
1 5H [(1 N
en t .e the tern

Q ' ad-wei lt loss if the
of the of collateral forfeited as a de g1
' e proportion

19

M . - . . .
entrepreneur has a Ship-wreck- The term (”Nay ‘ 1) glves the utlllty oftfie hg/z-abzlzty

in the case Of perfect information, SO in an asymmetric information case, 121:9 utility Is
educed t0 the extent of these ‘ Sorting COStSa.
r

Our next step is to Show that there exists a pooling contract which dominates this
separating equilibrium. We can then prove, along the lines of Hellwigl46], that the
sequential Solution to this game iS a POOIing equilibrium. A pooling contract does not

difie’entiate between entrepreneurs. We begin by conjecturing that this pooling contract
in"GIVES all entrepreneurs paying an average interest rate given by F = 7A tug-'1) A7}?
(We Will then prove in Proposition 4 that if this pooled contract Pareto dominates the

se arat' uilibriurn then this contract iS indeed the SOlution to this sequential game).
9 mg eq ,

The DRYOfis to the entrepreneurs under this p001€d contract are given by,

M ._
UL? : #(y — T)
. MH _
Um, = "Rf-(y ‘ r)

We See that the low ability entrepreneurs WOUId always prefer a pooled contract to the

nce of the high-ability t
separating ( ULL << Um), but the prefere ypes would deen d upon

the 80 t' g costs 8 Sorting COStS or collateral COStS in this economy are inve
1' 1n -

3913’ related
7 ° ' l econom . We also see that t
to «$9, , the degree of homogeneity In t 1e y he ave

. It!
rate 1‘ d r a pooled contraCt, ‘77, is Inversely related to ‘1‘-
11 e

AIH‘ (The Pooled int

ereSt rate is
8190 ‘ 1 t d t the proportion of low-ability borrowers which We
A l riversely [‘8 a e O ’7’ Slime to b
giver 1 large) Thus we see that given the parameters ,8 and 7, both S and p are inversely
‘73 S . '
relat Q

i i I 1i hthOllse Wthh lIlCrease
A, ltll the [OVlSlOIl Of a g

S tl'llS hOln - A1
0
geneity Flt,
a Separating equilibrium,
the § - ts are reduced Wider
*7? Q. 1.ng COS

but the aVerage interest

20

rate under the pooling COntr ac t is also reduced. The efficient en trap
renews

W17] linear
this sorting cost so long as they end up With a payoff which is more than the payofi" they

will get under a pooling Contract This defines 3 Cut Off 5": given by:

1+ 5'(9)

‘
\

MH_ (1.4)
V719)
With 5 > S", a pooling contract will Pareto dominate the separating contracts. This is
shown in the diagram below, where B1,, BH refer to the Zero Profit curve for the banks
on contracts for low ability and high ability. The dashed line B p refers to the Zero Profit
curve for the pooled contract. UL and UH refer to the indifference curves of the low
ability and high ability. (21,0) and (771,011) mfer t0 the eqmlibrium separating contracts
for the low-ability and the high-ability entrepreneurs 1‘ GSPGCtiver' The pooled zerO profit
line tor the bank, B p, lies below the indifference curve for the high-ability borrower'

The high-ability entrepreneurs“, in this case Will prefer a 1,001ng con tr act to the

Separating 0119- (1",c’) refers to one Of the multiple pooling Contracts which StriCtly

dominates the separating contracts (Th0) and (73>

C .
H) The p mOf below establishes
that among all Of the multiple POOled contracts WhiCh dominate the separati
11g contracts

(all Contracts like (7",6’) WhiCh lie below the indifference curve of the hi
ability U
. H
and above the dashed line 319, the zero profit line Of the banks for pooled CC)
0 I - tracts): the
n y I ooled contract which will be sustained as the equilibrium is the contr

t (F c \
O O C , \ 0
[Prog 36 follows along the lines of Hellng [46] and IS given in the appendix]. )

:Proposition 4 Given that S > 5* , in the three stage game considere d b
a 0216,
the sequential equilibrium 0f the game is given by the optimal pooling contra t
C

<ch:0>'

21

Figure 2; A Pooling Contract dominating the Separating Contracts

(PH

CH’)

 

 

 

 

 

The proof thus relies on the refinement of the game being walled in the credit market

lt'iple

It is because of the sequential nature of this three stage game, that out of the 1““

P001ing contracts which dominate the separating contract given above, there will perSlSt
, _ 6
only one pooling equilibrium given by (730)-
While the inefficient entrepreneur always prefers the p001ing contract .

“g compared
to th e separating contract, the efficient entrepreneur s preference will depend

 

on the co].
lateral costS- The intuition is that given a level of heterogeneity among e
. reprenelll‘s
1!)ch §§ing signalling costs will dissuade the effluent entrepreneurs from di
4———-——- , . . . Brentia '
6H§ 11“”qu mentions that this conclusion IS reversed if the Informed agents move first ting

following the

Cho - ' ‘ ' oweve , followin Wilson 69 w
EK§_ i KrePS sequential equilibrium. H T g l l, 6 can assume th s t l s r
1 "I" a e

stinguiSh among agents before these agents choose the contracts. This might also re fl t
ec 180
cal

un b

cons - ' b law that all a ents be iven the san . .
tr §ints, where 1'3 may required y g g ‘8 Opportunities. Therefore

the \111 i formed party (the banks) moving first is a plausible assumption.
n

22

themselves from the inefficient types. They are better off under a po oled/wza’zfierenaated
contract. Increasing provision of public gOOds in the economy redaces the [eve] thet-

erogeneity and thereby increases the Probability of a pooled contract dominating the

separating Contracts.

1.5 “Switch Point” and the targeting of public in-

puts

From (4) we can calculate the critical point at which the pooling contract dominates,
and the economy switches from a separating equilibrium to a pooling equilibrium- This

“switch point” is given by [Calculation in the appenle],

l—fiil; (1—7)(1 —fi)

_—
_—

(1.5)
1 _ ME. ’7
N

 

 

Where 7, 3 are the given parameters.

. . 77 ' ' t ' '
In Our analysis, the “5WltCh pOlnt lS Importan , because It IS a point 01: comparison,
local] th nefits of a lighthouse under a separating e(llll'libri'um
3*, between 6 be ”is a vis the

Wolf 11g eqvilibrium. Public goods which increase the homogeneity beyond

this switch
pain 1: results in the capital market having pooled contracts which dominate t

Contrfii ts A pooling equilibrium in this case, 18 soc1ally optimal because it .
c

creases the
utilitf f both types. We can compare the marginal benefits of both light-h0u8

(A) and
(B) a E this switch point defined above. This would tell us something about the

ChOlCe Of
the 1 i §htl10uses to be built.

Q—‘:Q:.mparing tllC P3y0ff8 ‘0 the low-ability types under a [”01ng and a separating

23

Marginal costs and
benefits of
lighthouse (A)

(MC = MB)

 
  

 

\

 

Li

 

 

fl

Lighthouse (A)
Separating

Equilibrium v Pooling Equilibrium

‘switch’ point

Figure 3; Targeted Public Input

equilibrium, with the provision of lighthouses of type (A): We find that tlle follOWlIlg

holds, [Proof in the appendix],

Proposition 5 At the switch point, the marginal benefits of light/muss /Aj
are higher under a separating equilibrium as compared to the pooling 89712725
flum- The marginal benefits of the lighthouse (A) are however, increasing

under the separating equilibrium till they reach the switch point. After this

the economy switches to a pooling
equilibT’lum and the marginal benefits of the lighthouse (A) then start falling.

We see in the figure above that the switch point separates the separating and POOIing
equilibria. The marginal benefits (MB) are increasing when separating equilibrium ex

24

(MC = MB)
Marginal costs and
benefits of
lighthouse (B)

 

 

 

 

 

»
~
.
.\
u
o

Pooling Equilibrium

Lighthouse (B)

Separating
Equilibrium

‘switch’ point

Figure 4: Pure Public Input

ists, and falling, when pooling equilibrium exists. Given a. constant marginal cost(MC) of
building the lighthouse q, the optimal amount of lighthouse (A) (where MB a MC) leads
to a. pooling equilibrium. On the other hand, doing a similar analysis of lighthouS (B)
e

we get the following[Proof in the appendix],

Proposition 6 At the switch point, the marginal benefits of the light/201155
(B) are higher under the separating equilibrium, as compared to the pooling
equilibrium. The marginal benefits of the lighthouse (B) are however, decreas-
ing both under the separating and pooling equilibria. However, the optimal

amount of lighthouse { B ) leads to a separating equilibrium.

In the figure above We see that in our model, to the left of the switch point we have

25

increased heterogeneity among agents, associated with a low level of public inputs in the
economy. The separating equilibrium dominates here. In this scenario, a public good like
lighthouse(A) which is valued mor e by the inefficient in the economy exhibits increasing
marginal returns. The reason being that the extent of private signalling through collateral
is high at this level and therefor0 greater benefits accrue to the efficient entrepreneur
through reduction in these private signalling COStS- Lighthouse (B) on the other hand,
benefits all entrepreneurs and reduces the heterogeneity in the economy to a lesser extent.
The indirect capital market effects in the case of lighthouse (B) therefore increase at a

decreasing rate.

as then suggest that given the a constant marginal cost (1 of
1—%“ <
building lighthouses - in an economy where the level of heterogeneity is high (1’ 15‘“

The above proposi tio

Lilli—3'1), the optimal choice would be building lighthouses 0f type (A)- The optimal
level of the lighthouses of type (A) take the economy beyond the SWitCh Point” Where
the pooled contract dominates. On the other hand, building lighthouses Of type (B)
would Inake the economy persist with separating contracts in the capital market. Thus,
considering the indirect capital market benefits ' we would always end ”p we}, pooling

EQUilibrium, if we build lighthouses of type (A). Therefore, when sorting COSts are high
in tt1 (3 credit market indicating a greater level of heterogeneity among borrowers, build-
ing I) uth inputs which benefit the inefficient results in the maximum capture of the
eXteI‘rl ality of the public input which accrue through the capital market. To the right of
the Switch point, under a pooling equilibrium however, we have to compare the benefitS

“0111 the two types again to make a choice. This is done in the following section.

26

1.5.1 Public inputs and the pooling equilibrium

Lighthouse A The total benefits 0f the lighthouse (A) are calculated as follows,

(i) “Direct” benefits to the low ability, given as 4'16. There are no direct benefits to the

high-ability, as before.
a? will

(ii) “Indirect” effects stemming frofr1 a reduction in the pooled interest rates 69,
benefit both types. Essentially, with the P1“ OVlSiOIl Of a PUbliC 800d which increases the

average ability of entrepreneurs in the economy, the risks associated with lending are

lowered, and this gets reflected in the lower interest rates.
Depending upon the

(iii) “Entry” effects in this case, also will benefit both types.
ates

Opportunity costs of the alternative foregone: With the reduction in the intete‘5t r

more entrepreneurs will enter the fraY-
(B)

Lighthouse B All these three effects are also reflected in the case of lighthouse
WhiCh benefits both types. Additionally, the high-ability also benefits from the direct

3—? ’ .
69’ Whmh Summarizes

productivity effects. We also see that the fall in the interest rate,
the capital market effect is greater in magnitude to bOth types in this case, as Cowpared
t0 the lighthouse (A). Thus in a regime where the pooling contract dominates, t here Will

be 3,11 unambiguous choice in favor of the lighthouse (B) by both types. The Same is

shown in the figure above, where there is a comparison of the benefits under the pooling
qulilibrium. (++ refers *0 bCHCfits of greater magnitude as compared to +), Thus to
the ri ght of the switch point, where the pooled contracts dominate in the capital markets

_ L1 1% Optimal choice of the type of lighthouse to be built is 8., those which benefit all

ent-1‘ epreneurs in the economy.

27

POOLFNG EQUILIBRIUM.

Light hOUSC( A) Li ghthouse(B)

 

A

  

‘Direct’ effect (+1 (+ ‘Direct’ effect (+)
Interest 1' ate effec ) Interest rate effect

 
   

 

 

   
 

8L
(++)
Interest rate eff“:t ‘Direct’ effect (+)
an (+) Interest rate effect

  

(++)

‘type’ Of Public inputs that Should

We have thus arrived at a, rough guideline fOI‘ the
hip between

n economy. The results suggest that there is a key relations

be provided in a
When the

ublic inputs and the degree of heterogeneity in the economy-

the type of p h. ve
to 33 19

heterogeneity in the economy is high, the transactions costs iDCUrred in order
g equilibrium are high. In such a scenario, a public input Which is hen-

the separatin
ability types results in higher benefits as compared to a PUblic

eficial only to the low-
good which benefits all types in the economY- 50, an economy should begin by buiI ding

PUblic inpUtS which are targeted towards benefiting the less—efficient in the ecOIIomy
' - ' ' loitation of the externality which accrue to th
ThIS will result in maximum exp e eCOnOmy

via the capital market, With such public inputs, the economy will end up haying a

p0019 (1 contracts which Pareto dominate the separating contracts in the credit market

After building a critical level of such public inputs, only then should public goods which

benefit all types should be built.7 In other words, if we explicitly take into consideration

7 ()ur result can also be seen to he in consonance with Boadway and Keen [20], Where using a different

‘28

the indirect benefits of PUbliC inputs via the Capital markets, those public inputs which
are targeted to benefiting the inefficient in the economy provide the greatest benefits.
Provision of public inputs which benefit all, or Which benefit only the efficient in the econ-

omy should be taken up only after buildlng a critical mass of such targeted public inputs.

1.6 Optimality Rule for lighthouses

This Section explores the optimality rules for the provision of the lighthouse using non-

dlStOI‘ting, lump-sum taxatiOILS WC Shall be considering the costs and benefits of the

lighthouse to the entrepreneurs under the two cases analyzed above (I) In the separating

equilibrium, and till where we have a pooling equilibrium.

(1) Effect of a lighthouse. on the Separating Equilibrium: The Optimal provision of light‘
houses oi type (A) will never result in the separating equilibrium, therefore the cost-
benefit, analysis in this case will be restricted to lighthouses of type (B) , Where the

marginal benefits are declining. The government maximizes the Welfare (W) amo h
ngt e

entrepreneurs, which is given as

Alan: W = "LULL + "HUHH + nb7rb - T
9(8)

 

 

\ffir
’nfOI‘I‘rlation structure than ours, they conclude that there should be a conventional “over-supply” of

publi Q goods which are more valuable to the less efficient, and a conventional “under-supply” of DUbl‘
‘ 1c

gquS which are valuable to the efficient.

a. Si nee labOr is non-elastically supplied in this model, it does not require any additional assumptiOHS

29

nL,nH and Nb refer to the nurnber of low ability, high ability entrepreneurs and the
number of banks respectiveIY- 7th is the profits of the banks, which is zero in equilibrium.
W is maximized s.t. the Government Budget constraint, where q refers to the cost of
the lighthouseg.

W = mesh-Aggy — 11+ "H(98)[%y ‘ 1 - CH0 - fiXLNMHH — qu (1.6)

Maximizing W with respect to g we get the following,

6W 1 — [3) 66 N — M
5;; Q Era-(1%) + ””[CH—(T — 39in — 'B)(_N—£)] + ”’LCQBXULL) + nh(93)(UHH)°

Enitjt) are the “direct” productivity effects referred to in the Samuelson Rule as EMRSy'9°

I
(t' ' 3) . iS
The term in the square brackets refers to the indirect capital market effects. The n

refer to the “entry” effects.

(11) In the case oi the pooling equilibrium, ‘g’ refers to both types of lighthouses, (A)

and (B), The government maximizes welfare (W) with respect to g,

w = 2ni(g)[%(y-T)l+nm—T

s.t. qg = T

Maximizing W with respect to g and the government budget constraint we get

g; = Eni(g)[% — %(%)l + Eni(g)Uii~ (1.7)
fl are the direct productivity effects Which can referred to as ZMRSW. The indire0t
CRID‘I t. al market effects are seen through a reduction in the pooled interest rateS, 01; being
negative. The ”:5 refer as before, to the “entry” effects. Using the above notational

9T1“; lighthouse is assumed to he built at a cost q, which is given ex-ante.

30

conveniences, the effect Of a lighthouse "1 ”“5 economy can be summarized as,

q :2: ZMRSy,g + ‘Capital market’ effect + ‘entry’ effect-

1.7 Conclusion

This paper makes two contributionS— F irstb’, it analyzes the role of public inputs from
an information theoretic approach- PUbliC sector economics is increasingly realizing the

importance of imperfect information as a constraint on public policy. Private (asymmet-

ric) information limits the set Of allocations that can be achieved in an economy While

is has been

mation asymmetry between the government and individua

the nature of infor
xist ammo

a

traces the effect of public inputs when asymmetries e

analyzedl‘lfll, this paper

individual transactors. Niarket allocations in this second best WOrld are often

Contracts which are designed to prevent the inefficient from mimicking the efficient (SCI)-

arating equilibrium), or by Contracts WhiCh go by the average attributes in the economy

(POOIing equilibrium). When separating equilibrium exists, one of the tests of an ofiective
PUblic policy is its ability to relax the self-selection constraints. We Show in ear mOdeI
that a pUblic intermediate 800d, by reducing the heterogeneity among borrowers is able
to rel ax this self-selection constraint in the credit market. Those public inputs which are
better equipped to reduce this heterogeneity have higher indirect spill—overs. TherefOre,
Wile 11 the level of public goods in the economy is low, our model predicts that public

goods which are valued more by the inefficient. (lighthousc(A)), should be provided, An

Optitnal provision of Such targeted public inputs results in the pooled contracts domi-

31

nating the separating contracts in the credit market. Here, the targeted public input
dominates the pure public input. In other words, it makes sense to restrict access to
public inputs when the marginal (303‘? Of another user is zero. When the credit markets
are characterized by the pooling equilibrium, public inputs of both types, improve the
average, (in this case the average quality of loans made), and encourage further lending
in the economy. Public intermediate gOOdS can thus be perceived as “public signals
which determine the level of socially Optimal investment.

Secondly, in the light of the above, this paper calls for a. rte-assessment of the Cost-
Benefit an a Iysis ofpublic inputs. Infrastructure and other public inputs can be justified
without reference to the credit market — and yet, we have seen above that the indirect

SDill-overs from the credit market are too large to be ignored. This result has key

policy implications in determining the Optimal level of public intermediate goods in any

economy. One has to go beyond ZMRSy,g given by the samflelson rule in the case Of
public inputs. This is especially crucial since private investment decisions cannot be
de—linked from the outcomes in the credit market.

There is scope for further work in this area. The information asymmetry in t he Cred'

1t
mar ket is ex-ante, giving rise to the problem of adverse selection. An extension to this
mo (1631 Would be considering the effect of public inputs when the credit market is also
Plag ued by moral hazard issues. Secondly, the taxes considered in this mOdel Were n0n_
diStQItionar-y. Further work can also be done on the nature of distortionary

taxes and
their effect on the marginal factor cost of building this lighthouse.

This analysis also has implications for real world public policy, especially, government

901 icies in the credit Inarket. Much of the analysis on public interventions in the credit

32

market veers around the Optimal level Of SUbSidieS and loan guarantees to lenders/61, 38/
or redistributive policieslloi 49] ’ WhiCh improve Credit allocations. This Paper on the
other hand, deals with public pI‘OViSiOn 0f baSiC inputs like infrastructure, health, literacy
and other services and its effect 011 the credit market. The main feature of such public
inputs is that an increase in their PFOViSion leads to a narrowing of the spread of abilities
among agents. We show that it is preCisely this feature of public inputs which relaxes the
Self-selection constraints and achieves Pareto optimal improvements in credit allocation.
This feature of the public inputs also allows us to make conclusions about the targeting of
PUblic goods to specific sections of the economy. There is thus a need to carry out more
dis‘Segregated analysis of public sector capital. As we have shown, the effects of public
inputs on the economy differ, depending on the type 0f the public input considered.
M ()re importantly, it is seen that pilblie expenditures on infrastructure and provision
OfserVices like education and health are normally justified on grounds of equity as being
eXpellditures on ‘merit’ goods, or as interventions in the production of goods having
Signifi cant positive cxtcrnalitics. This analysiS points out to another equally important
lUStifi cation: public inputs, by equalizing ex-ante abilities among agents can relax the
self‘ :3 Qlection constraints and improve the efficiency of market allocations. The ability of
vari C) \1 5 public inputs to achieve this slack in the incentive constraints differs. In the light

of t1”1 ‘1 result both the quantum of government expenditures, and the areas in which it
s k 1 .

18 b Q i ng Spent, needs to be re-examined.

33

LA Appendix
I.Proof of Proposition 2:

The problem for the bank is,

w WULL + (1‘7)UHH s.t.

{n- .63}

 

(i) the Zero profit constraint for the banks; [It—5m + (1 _ 1:9)561' = 1.

(ii) the self selection constraint ULL 2 UL”.

The usual strategy is conjecturing that the self selection constraint which is binding
is ULL = UL”. We later verify that the optimal solution does satisfy the second self

SEIectiOH constraint U H H > U H L. Substituting (i) into the utility function subject to (ii)

3

~——

%L=Wl%‘(y~fi—fiCL(1—MNI))‘CL(1‘%L)-ZL]+(1-7)[Mn( N

(ch? CL} N M”
flc”(l — 7%)) — CHU ‘ l‘72“) - an +Al7v"(y“ “AT; — 30111 - 1%)) -014“ _ 1173‘) _ Zr] _
M

[#(y __ 31;: _ecH(1_MUZ))—cH(1-%h)-Zrl.

 

WherQ A is the lagrangian multiplier associated with (ii). Setting 3L
CH

6 = 0 gives us,

(1 — 7)[(1— KNEW? - 1)]
= “(l-flexed) >0'

Whi Q}‘l implies that (ii) is binding. Using the above result, it is easy to show that 51’ < 0
CL

’

. _ , - - ' 3 _ lV -
WillQll ifllphes tklat CL : 0 Putting thlS Into (1), we gCt TL — m. Pllttlng these results

in (i i ) We get,
1 “(919711)

1_ML
N

CH =
We & lestitllte for r” from (i), and solve for the same above to get the Optimal value 6-H.

34

We get,

A!
., _IL‘NQ 172‘)

(3” .,
It can easily be verified that U” H > UHL’ and that our conjecture was right.
II.Proof of Proposition 3:

The benefits to the high ability through lighthouse (A) is given by,

3UHH
a9(A)

N- NM”

1’ {(81:11, )[(1— —fl)(——— ———)] (1.8)

USjng the definition of CH given in Proposition 2, we can write,

 

36H NMH(1‘fl)(MH ~N)
aML = [MH(N — ML) — HAM (1.9)

We See that (1.9) is negative since (Mr! —‘ N) is negative. Therefore, (8) is positive. The

benefi ts to the high ability through lighthouse (B) is given by,

BUHH y (1 — )8) _ 36” + 8C” _ N -
55“) N N H 8M1, 6M” )(

 

 

 

Mflll- (1.10)

6C .
Frorrl (1.9) we know thatd iti— is negative. 48;, is then given by,

 

 

00H NMLU — .3)(N — 1‘4le
W}; = MN — ML) — HMLUV — Mm]? (1‘11)

(1.1 l D is positive. Therefore, the condition for Proposition (3) to hold is that,
BC 06”
I .” l > |—-—l
BAIL 81W”
2:» MH(N -' JWH) > MLUV ‘ ML)

ME, — Mg
1V1” — ML

 

=>1V > A!” + NIL

 

Which is true as per our 35301111) tion.
III.Proof of PI‘OPOSit'iOn 4:
Suppose bank j deviates from the pooling Contract (7,0) and offers another contract
(F75) _ it will be accepted by all the high ability and not the low ability. Since the
high ability types are more likely to accept a contract involving collateral to signal
their ability. Consequently bank l 1' eceives applications from all the above average, high
ability entrepreneurs. Banks that have Offered (F, 0), will then be left with below average
Sample of the population. Since ('7‘, 0) only breaks even at the population average, all
applications to (710) will be rejected in Stage III. Knowing these rejection strategies, all
entrepreneurs, both efficient and inefficient, will apply 0111)’ for (fl 5). This is contrary to
Our Earlier assumption that only the high-ability Will Emmy for this contract. Thus, under
equl'libr-mm strategies, deviations to the optimal pooling contract will be rejected and it
0811110 1: be upset by any separating Contract (7, 5)- Therefore, (F: 0) Will be the equilibrium
13001ng contract, and the sequential solution to this 3 stage game considered.
IV-Q Qlculation of the ‘Switch P0int"

A8 p Q :r (1 .5), the definition of the switch point, we have

N+cH(1—fi)(N-MH) = MH": (1.12)

USi ‘ ' - —- ____———N—————'- we have
11 g the definition, r : IWH“7(1”H”AIL),

N7(MH — ML)
MH — "((MH — ML)

 

CH(1 - fixN " NIH)

MAM—ML) we have

Us‘ - - - . —
111 g the definition, (.11 : m—BMMN—MH)’

(1 — [3)(N - MH) “/

[VIM/V _WBMAN - 1W”) — MH — “/‘(MH - ATL)

 

36

 

 

Which gives us,

(1 —mNMH — (1 — BMNMH 4" (1‘ WNML ~<1— 5W}. +(1~fl)7A//2
[I

And this implies,

ll

mm) Mm -' W1 ~ 7)) — NMH<(1— mu — 7)

=> ML _ MH[MH(1‘5)(1 —7) -+:7.:\fv(v—-(1—B)(1—~)))J

:> A41, = MH[1— MW

-7)

 

 

N 7 J
1_LWJ._
1—L‘1d — “r
N

-Pl~oof of Proposition 5:
At the switch point, the paYOffS '30 the high-ability entreprenems With a separating
Contb #3013 just equal the payoffs they would get With an undiffer entiated/ pooled contract
This payOfi is defined as per equation (1'12)'

sepQrating Equilibrium:- For the type aL entrepreneur, it is easy to see the m . l
argina

beYLQ fit of the light house (A),

BULL _ aULL _ 3/

89M) _ (9le _ N (1-13)

 

 

F .
1‘0 11‘ (1.8) and (1.9), we can calculate the marglnal benefit of lighthouse (A) to the type

a” Q neur
_ lltrepi‘e 7

BUHH _ BUM” A/IH(1_/3)2(N "11”!!!)2

fl‘

05m) ’ BML [MH(N—ML)—[3ML(N_ W (1.14)

 

37

 

The marginal benefits of the lig’;hthouse (A) tO this economy as a whole under a sep ting
313

equilibrium regime are thus given by (1'13) + (1-14)- We see that the marginal benefits
W > O.

are increasing in gm), Slnce BA! 1.

Pooling equilibrium:- At this point, the Inarginal benefit of the lighthouse (A) to the

low—ability entrepreneur is given by,

 

 

BULWZQQEZ:£_ (1—7lMH 115
89(A) aML N M71! — 7(MH - MLll2 ( ' )

The marginal benefit of the lighthouse (A) at this point to the high-ability entrepreneur

is given by,

8U 8 UH ”Y My
H7 7 __ x (1.16)
a9(A)

 

.—

 

aML ’ [My — «My — Map
(1-15) + (1.16) gives the total benefit of the lighthouse (A) t0 the economy under a
90011 11g equilibrium regime. It is however decreasing in 9H) Since both $1321 and 952%}?-
are '< o.
(”HIS the marginal benefits of the lighthouse (A) are increasing under a separating equi-
libri Kim , and they are decreasing under a pooling equilibrium. We now have to show that

at the switch point, defined by (1.12), the marginal benefits of the lighthouse (A) under

“1% rating equilibrium exceed the marginal benefits under the pooling GQUilibrium.

$6193

At L118 switch point, given by (1.12), “assume true” that (1.13) + (1.14) > (1_15)+ (1.16),

 

 

 

Z, MH<1—.6>2(N —- M”)? 2 > ., mm ‘L
lAIHUV —~ AIL) — [3A,],(N - NIH” [1)!” — 7(MH _ AIL“?
(l—fi)(N-MH) (27—1)

 

_________/ > . fix—
:> [MH(N — A41.)— fiMLUV — MHll W" ‘ “M” “ ML”

 

N(A’IH—A’IL) . 1_—. : 1V 1 t;
“ - ,an(I__——\___._weiaveo
Usl I 1 g the defimtlomi CH : MMN—Aim-aileN—Mn) Mil-*erML)’

38

Show,

CH(1 --fi)(N— NIH) rm
W > T

We now use the definition 0f the SWitCh point given by (1.12), and we have,

MHFxN > F,/27—1(MH—ML)

=>F(MH‘ WOW” ‘Md > N

Again using the definition of ? given above, we have to show,

N(MH - «27 —1(MH —’ MLfl > N
My —7(MH - ML)

Since ’1 < 1 and 7 > m > 0 for all 7 7g 1, we therefore have,

(Ii/[H — V27 —1(MH _ AM
My —’)’(MH ‘ ML)

 

The Inarginal benefit of the lighthouse (A) at the SWitCh point is greater under the
sepaatating equilibrium compared to the pooling equilibrium.
VI - broof of Proposition 6:

SGDQ Ta ting Equilibrium: The marginal benefits of lighthouse (B) to the inefficient is given

by

\- 3

aULL 3/

— — 1.17
(99(3) N ( )

 

Frq) 1*) the proof of Proposition 3, we get the marginal benefits of the lighthouse (B) to
1

 

 

 

thq} (efficient entreprei’ieurs,
'» — ' v N ~ A!
()UIIH y (1 (3) _ LC” + dc” 1—fi) —\fl . (1.18)
69(3) — N + N C” (BML 8114”)“ ( N )1

39

 

 

 

 

 

 

 

 

 

 

 

(1.17) + (1.18) give the marginal benefits 0f the lighthouse (B) ‘30 this even
' omy as a

whole under a separating equilibrlum regime.

Pooling Equilibrium: The marglna1 benefits of lighthouse (B) to the inefficient under the

POOIing equilibrium is given by,

 

 

6UL'7 : Ej—\ A—A/MA’IH-ML) 1
89(8) N [Ii/[H - 7([1/111 — NILHQ ( 19)

The marginal benefits of lighthouse (B) to the efficient under the pooling equilibrium is

given by,

aUH'y _ EL+ (7)(MH-‘ML)
‘ N [NIH—7(MH—W (1.20)

 

 

59(3)

(1.19) + (120) gives the total benefit of the lighthouse (B) to the economy under a
p001ing equilibrium regime.

We {3 Ist show that at, the switch point, defined by (1.12), the marginal benefits of the

Ugh t house (B) under the Separating equilibrium exceed the marginal benefits under the

p 001 i ng equilibrium.

At t he switch point (1.12), “assume true” that (1.17) + (1.18) > (1.19) + (1_20)_

 

(1 — We”) __ {99; + 29.44“ ‘WN ’ MM} > W
N 8ML aMH N [Mu — 7(MH _ Mn]?

 

Usih g (1.9) and (1.11), we can rewrite (1.17) + (1.18), and we therefore have to show

1*, . 1—. 2N—MH)[MH(N-MH)-ML(1V-MLfl
“#1) ‘1' {( fl) [(111]!(N—A’IL)‘B/UL(N’1”H)l2 }

> (Mu—MLXQ‘r-ll
[NIH—WM'IH—IVLHQ

UR ‘ - - .— M m .
“ 1 1%)?)- the dfi‘fimtlon, CI! = M”(N__ML)-BML(N—MH)‘ “'9 have to show,

l—f3)((f (1—[3)2(CH)2(N—A1H)|AIH(N-AI”)—A’IL(N—AJI_L)1
L N "—1) + i [N(MH—ML)]2 }

> (Mn-4400741
[Ilfu—~1(1\1H-ML)12

 

40

, ' . —_ , ) 1V“IUH—AIL
n-mc ) 0—9)) CHW" M”, > \M
:> g H { (N(MH"1‘!L )1 } [1"IH-W(MH-ML)1

NUUH—ML)

N
. . - '1 —— A c
Usmg the definition of 7 =‘— NH , ’7UUH‘AIL) ’ We have to Show,

 

 

,,_ 2
<1 — .B>CH[N(MH —- MM + (1 '5) “WW - MHHN — My — ML} (27 —1)r2
[NUVJH -— Md]? } > T
At the switch point(1.12), we have (1 ‘ ’8)CH = W, we therefore have to Show,
(MH? - N M _
N _MH)[N(MH__A,[L)]+ N I AI], [N—A/IH _AlL] > T2(2’7—1)(A4H—1l/IL)2

In the proof to proposition 5, we have Shown that,
Mm: — N > r\/2v - 1 ( My — ML)

Therefore, in order to prove propositiOn 6, it SUffices to show that,

M % _
WWW” ~ ML)] + (My? — NW ~ MH - MI} > (Mm — N)?

\MH

:> N(MH __ N1134— (1wH7— N)(N— (WM) — (MHF‘NNWL > (fl/IHF~N)(N— A’IH)

=> N(MH — M.) > (Mu? - NW1.

:NMH > NIHML?

ML
7A’1L +(1— 7')MH

 

:>NA’IH > NMH

\ ~ .
V5 1 ch IS true because,

AIL
71"IL +(1— ’7)A[H

 

< 1

It i & thus proved that at the switch point, the Marginal benefits of the lighthouse (B)

are: mating equilibrium as compared to the pooling- equilibrium I

greater under the sep

41

We now show that the Marginal benefits 0f the 1ighthOUse (B) are de
creasing bot/2 under

the separating equilibrium and Under the pooling equilibrium

Separating Equilibm‘m" Refer “'17) + 0-18) as MBS the Marginal B fit {1' ht
v ene s 0 1g —
house (B) under the separating equilibrium.

 

itIBS=2_iZ+(—L—%)m+{( )[(AZBMH)1MH(N*MH)-ML(N—ML)l
N ”(N B ML) ‘ BMLUV — MH)l2 }
(1.21)

We have to show that 95%;? = 23%? + %’5 < 0.
T0 Simplify the calculations, denote the t<9I‘In 1'11 the denominator, [MH(N — ML) "
flMLUV-MW/ as 45 and the tel“1min the numerator (1 “ 3)2(N ~ MH)[MH(N ’ M”) ..
ML(N — ML“ as Q.

Thus, MBS can be \Nritten as,

_ 3g (1—5)(CH) 9
M85 — N+_1\T\+ 53

Which implies that ,

8MBS _ (1 — [3) ac” + {W1 — WW — Mn)(2ML ~ N

 

 

 

 

aML “ N all/IL @4)+92¢(6(N\ M
”+11;
11)]
m — N BMH + W )
L /
{MN —' MHXN — QMHH — QQ‘MN - M }+

 

 

 

(1)4 w

(1.23)

42

 

 

 

Adding (1.22) and (1.23), and collecting terms, we simplify the Semein th [/1 "mg
(5’ 0 0W1

way:-
The coefficient of the term (1’2 ( 1 “ (3)2 is,

{(N — MHMN 1M”) — (N .. 2M5); ~

 

 

LMiW — MH) — MLtN — M )J}

4
V

. — , 5110“,: +
Which we see 18 less than LCI'O. To that, We see that the (+) term has been proved

. . - d the - .
to be POSitiVe 1n Pmposmon 3’ an ( ) term 13 negative because 2M3 > 2M L-

Take the coefficient of the term 324) whiCh is, {13(N — MH) + My — (N - ML) " 5%} -

v

This term is less than zero because: {(1 _ flMMH + M L — N ]} is less than zero.

- , ‘ (1‘5) Bea C r
The third term to be conSIdeled 15’ { N {IBM}, —+— %fi:]}. This term is less than be“)

ac . t
because it follows from“) [fi%1< O and [51%] > O, and we showed in proposithn 3 tha ,
| 86” l > I BC” 1
aML 8M”

‘ 4 . 't '8 shown that 6MB .
And the denomma‘OT» ‘1’ > 0 Thus 1 l 35532 15 less than zero.
Pooling equilibrium: Denote (1.19) + (1.20) as MBP,

the m -
argmai benefits of the light-
house (B) under the pooling equilibrium. It is given as,

 

__ 2y (Mu - ML)(2’7 ~ 1)
MBP — N + [MH‘7(MH_TL)]2

(1 24
61x18 ) aAlB - M . . )
We see that 5111‘ < 0 and '45!” < 0. Therefore, d913,

 

 

LB Notes to Calculations:-

(I) The Perfect Information Equilibrium.

The problem is,

 

 

aqu—Mi(y—T)‘—(1_ Mi)Ci\Z M M
‘{r. a}, u —’ N N i- S.t. N 7"- + (1 — thflq z: 1 (1.25)

Plugging the Zero Profit constraint Of the bank into the Utility function and differenti

ating it w.r.t. c,- gives us,

3U“ :

6c,-

M,
(1‘ y) (B — 1).

Which is negative, since fl is less than one. TherefOI‘E, 8-8 per the Kuh Tu k r conditions,
II- C 3

c1 = 0. Putting this illto the Zero PTOfit condition Of the banks w t h 'nterest rates,
v ege t e 1

Ti 2 £7, as the recipro cal of the entrepreneur’s ability.

(11) Effect oi \‘ighthouses on collateral: From Proposition 2 't
a we can re—WII 6 CH as,

 

 

 

 

 

C ::= N(MH — ML)
H MH(N — ML) “ 3M
__a_C_H_ _____ NMH(1-l6)(MH —N)
3ML [MH(N — ML) - BMW

We See that it is negative since (1W H — N) is negative.

 

 

9511. = ”ML“ ‘5)(N-ML)
8M}; [lWH(N - NIL) - [3&le
W11 i ch is positive.
44

 

 

Chapter 2

Pubh’c versus Private Signals in the

Credit M arket

2 .1 Introduction

The level of investment in any economy,

a large extent on the business environment in that economy. A Coun

tr .
. ° ' WbICh has
a poor implementation of laws and regulations governing Its bUSiness 00111:
l‘

. . ts, a 12'
level of bureaucratic inefficiency and COIIUPUOU, Poor quality of DUblic set 1gb

Vi
infrastrucwrea political instability etc. ..will not attract a high level Ofpriva

. e flourishing of private initiatiVes
Y1 environment conduCive to th Deeds t

obe f0

11 the b Stered
- - ' we » - . .
bot 1,1 by the institutions Wlthm an economy as pu 1‘0 pOllCleS Wh

lCh shape it
Tug 0 taken together, may be said to constitute what is today Called
‘ t“'

the u

1 cent years there has been an increasmg availability of empirical
~ re ‘ i
ll

45

M

 
 

 

 

subject} Based on such data, several cross-country studies find that 5/1089 was
C011!)

. . j - .
which rank low on this 500131 nfrastructure index (countries Plagued by corruption,

. ' I) ~ . . .
predatory busmess practlcesa a d rent seekmg actiVities) are also Stuck with 10W [eveIS

of investment, lower product1v1ty and l()Wer levels of income and growth. Mauro[56],

' tin o . . , .
for example, uses a data set CODSIS g f SubJECtlve indices of bureaucratic honesty and

. . ct of . _ .
efficiency to find a negamve imp a. Corruptlon on investment levels. According to

' a 0n . . . . . -
him, if BangladeSh were to achieve e Standard deVIatlon increase in its bureaucratic

efiiciency (which would take to the level 0f bur eaucratic efficiency in Uruguay), its in—

vestment rate would rise by five percentage POintS and its Yearly GDP growth would rise
by half a percentage point. Hall and Jones[45], conclude that Countries which achieve
high rates of investment in phySical and human CatlDital and thereby high rates of prOduc'
t'wity, consistently score high on the social infrastructure index. Recent Studies in the
- - ‘ 0 indicate that wide spread r .
”ammo“ economies\25l, als p edatlon and insecure Property
rights have depressed capital accumulation in these countries in all it .
_ . . . S duneneions. And
lastly 811,111ng WeilGSl, arrives at a Similar conclusmn With TeSpect t
, o
. . I‘ej .
investments in the developing economies. gn dlrect

Based on the above empirical results, the object of this paper is to pr
0

Did

e

u - . ‘ 1 ' f astructure indicators and inv a DOSS'b

m1 Cro” link between soc1a in r estment (1801'st 1 1e
8. 18h

by 6

go i 11130 the theoretical reasons explaining corruption or bribery undertake
n

"“8 (See

Bartlhanlfll for an excellent revieW), but given that varying levels of social infrast
ru

 

1 $118 laLBSt being an extensive Survey by the World Bank Group [70]

courl‘ '09 between late 1999 and mid-2000, on the business environment fan
_7 r1 ,.

ing firms) C;

Bug i it ‘84:; Environment Survey, (W BES)» 2000-

46

 

 

 

. . . unfit)" to countr ‘ -
exrst (which differ from 00 Y, or even differ across reglon . .
S Withlh the same

country), I would like to explain their relation to the borrowing and investment decisions

of firms.

' id) is iv - .
A common explanation Wh g en 18 that investing in such regions is generally

. ‘ ' ' . . 1') Wh . .
risky. This generalization IS t1” ”‘3’ en we conSIder that lenders (banks) can deVise

_ risk -
contracts which can separate the y prOJeCtS from the not so risky. Therefore, the

. is
Question I will be trying to answer that Why Would a lender (presumably, a bank in

. _ i ' - - .
this case) look to these economy W de mdicators 1nstead of firm-specific indicators 1n

taking its lending decisions?
The mainstay of the explanation Will be the existence of asymmetric information
in the credit market- I essentially bUild 0“ tWO models, (a) Hellwig[461a which takes

into consideration ex— ante, pre-contractual adverse selection PTOblems in a. three stage

dynamic game being played between the lenders and borrowerS in the credit market

l

and (b) Aghion and Bommlll, which deals with post-contractu a1 mo 1 h
ra

azardissue '
h dmak E' mod h hwnho ' Sn
t e cre it r et. arlier els ave s o W Imperfect infor t‘ 1

111a 1

on.

. . . I); .
market may give rise to credit rationing [65]. However, incrEflSing the the credit
C011

the banks to include both the interest and the collateral all 0We d for th t Space of
e

th ' l self-selection device and a ‘ - possibil'
e banks usmg collatera as a VOIdmg the prob] Ity of
e

sel . 16] In this aper I allow banks to simultaneous] of advers
es;-

COI ‘ b t ' ' ddition to the Dre-contr - rates and
lateral requirements, I! In a aCtual mfomlation
. In
MOI the lines of Stiglitz and Weiss [?], I add another level of inf Gt
l g 01‘ Ination aSYIn
In t. 1, 16 form of a post-contractual, moral hazard problem. r

This paper then shows that due to this interaction between SGleCtio d .
we

47

 

 

 

 

effects, private signalling loses its rdevance as a Self-selection device. Lenders would then
increasingly resort ‘0 imbue Signals like social infrastructure indicators. It is through this
link that such public indicators Which define the business environment of a firm affect
their investment decisions. I then go on to show that credit rationing could be more severe
in economies having a poor SOCial'infraStructure. Thus, in a simple theoretical model
WhiCh explicitly takes into accoth a parameter of social infrastructure (the possibility
that because 0f I300r implementation of laws, Corruption, need for ‘irregular payments’
to bureaucrats etc... firms do not earn their full revenue, which they would under ideal
circumstances), I Show how economies ranking low on the corruption index have higher
levels of investment. rThis WOU1C1 also explain, for example, given the global nature 0f
investment portfolios, there iS an increasing emphasis on ColleCting and understanding
country risk measurES by institutional investors.
The paper is organized as follows, section 2 introduces the model

Section 3 estab-

lishes the key result of the existence of a pooling equilibriUm in the Cred '1;
1 mark .
' - 6t. Sec
4 discusses the effect of social Infrastructure on this equflibrium and h tion
: t e

given in section 5.

2- 2 Model

A9672 t5 ._ Identical agents who are endowed with 1 unit of labor (l) Wh‘
. , ’ lch they
”18:1 .\ .tically. They are also endowed With wealth (w < 1),
i . Clin
bar 1 k or used as collateral to obtaln loans from the bank- The agents a
’ are risk.“e
. lit
11v“ ral,

for 2 periods and they produce 111 the first pcrlod and consunm in th
(Esccorul 7F}
' lose

48

If

agents have 2 choiceSI to be workers in a routine activity WhiCh 0111}, reqm'res ] b (and
a 01'
thus dePOSit w in the bankS) a or to undertake an entrepreneurial Project which requires
. - bo ' - .
them to combine 1 null; 0f 1a 1‘ Wlth 1 unit of capital. Workers get a deterministic

return Z. Potential entrepreneurs have to borrow 1 unit of capital in order to undertake

the project. More about the entrepreneurial preject will be said below.
Banks Banks in this econonly are riSk‘neUtl‘al and act as Betrand Competitors in
a market where they obtain elaSt’icaHy SUPpIied fLlnds. We normalize the gross deposit
rate at which the banks Obtain theSe funds to be one. Since we assume banks to be
Betrand competitors, they make zero profits in equilibrium on the projects they lend-

Projects : The rel: urns from the project are Uncertain. The uncertainty in this Proled’

stems from two sources,

Ecr- Ante _ The pig-xgct has a success probability of p. This is however not known ex.

ante either to the potential entrepreneur or to the banker' What the entrepreneur does

know ex-ante is the return from the project, if it is successfu1_ It is i d
n eter . .
° . 111111111
returns from a successful pI‘OJeCt that this model takes into the SociaI . g the
Infr
the economy. In an ideal situation, a successful project win .Yiel d the ent tructure of
T e

e
u " . e .
Yield (YF)- However, firms lose revenue in a business environment whj h at Its fu 11
c .
1

. - , OHS - .
and predatOTY- 1 base this observation on the World Bank Business Environ trzctwe
e
nt Sufi/9y

2000 (“.IBES). Respondents in this survey were asked if it Was Common f

' Or firm .

. . ‘ . . 8 Lin .
hue. f business to have to pay some Irregular additional payments’ to the"
0 get things (ion 7

e . 77

In %outh Asia and Developing East ASia more than 60% 0f the firms ‘d
831 this
Was

alx ‘\
V‘Ri.

that quell payments were at least frequently required. In the transition 9
‘ ' ,conomie,g
L Of

yb‘ mostly or frequently the case. In Africa, more than half of th e fi
’ rms reDort
ed

49

 

 

Central and Eastern Europe, around a third of the firms provided such responses. Only
in the OECD countries and the industrialized East Asia could this response be described
WBES

as rare - around 12% of the firms. To gauge the actual impact of such payments:

. n blic
enquired about the total percentage of revenues paid as “Unofficial payments to pu
of Eastern and

officials. These payments are the highest in the transition economies

dia, Pakistan

Central Europe at 5.5% of revenuesa in South Asia (which consists of 111
of the total

and Bangladesh) it was 5%, While in developing East Asia, it was 45%
and 83%

revenues, In contr 3315, 86.3% of the firms in the newly industrialized East ASia’
b ibeS lE’l

of the firms in the OECD countries reported paying 0% of their revenue3 in I

Therefore in this model, ICt the Proportion of firms who have to Sufie

on account of such payments be ’7- Let this lower yield be denoted Y1,

efilisi bUl»

potentially sufiet a range of lOW yields on this project on account of such p33

3L whethel

1 summarize .1" to be Y1” and YL < YF). Since such Payments are unoffici

their yield is YF 0‘ YL is the Private information of the firms -
' 7 ‘3 thus an indicator of
social inirastructure, which is public information. Higher the 7 hi gher .
, is t h
e pro '
of firms who have to lose revenue on account of factors such as Corrupt' portion
ion
a l) -
Fiberx 10b-

by ing for award of government contracts, payments for prOCUrement f
o

. - - ; , IiC Serv.
etc - .. and greater IS the Size of thls shadow or unofficial economy? 1063

2This assumption implies that firms who choose to make such payments are 1 t
‘10 cnga
. ged in .
tion ’(where predation includes rent-seeking and dupe activities [25D Therefore it is preda»
. , ve d.
t1] \ i h . -. . - ry lfierent from
L. Murphy, be leifcr and Vlblmy hypothcmb [bl where firms choose between product‘ (1
‘ We an Preda
- . tory
3C1: lV’itiee depending upon the returns guaranteed to them by the system. Here we haw ' 1
e a. Sing 8 produC-
lcct, but on account of the poor social infi'astmcturc, some firms have to m '
'dkc illegal paymL
)nts

tiv ‘
I“ pro

to i i I i ‘
Qarr out the pro Ject. This assumption lb Closer to the CFOSSI'na ' ? '
y n and Kim [J hypotheSis where some

50

 

Ex’POSt -' In the pOSt-contractual ex-post scenario, the success probability (I?) get-9 de—

. they
termined. p is an indicator of the effort put in by both types of entrepreneurs: once

1y
- - ° hosen on
secure a loan contract and have to implement the pI‘OJeCt. It IS optimally C

‘ ' tion
. . an lIldlca
after acceptance Of the 10311 contract. SlnCe the probability of success p IS

1] ,
. . .nd Boltoni
of the the individual’s effort, there IS an effort Cost (3(1)) Following Aghion a

we assume a uniform convex cost funCtlon across individuals,
2
A 4 6/19): 22—.

The Project is high yielding. We exPhCitly assume ‘high yielding’ to be»

mm the pm]
A 5 Y,- — 1 > Z, Vz' e {F, L} - The expected value 0f the net retumfif

exceed the opportunity cost 0f the PTOject to all agents.

Thus there exist incentives for all agents to undertake the pro j eet.

2.3 Capitafi Market Equilibrium

Taking, into consideration, this two-layered information asymmetry, the Q

. . apital market
e(allilibrium is obtained In the followrng way,

Th e optimal loan contract: Assume there exists only 'a standard debt;

. contract for
investing funds (issuing equity is costly). As in the model of Bester [16], banks ofl‘
81' a

loan contra ct consisting of a pair (7', c) where r IS the gross interest rate Charged
l aJId c
V . .
firlh§ are moral and WOllld nOi; engage ‘n COI‘TUDL practices, and some firms are amoral and would eh
. 7 ' . gage

 

 

in Qofluption. In this model however, the morality of firms is not fixed or given, but, it is a function of
the 1)aram0t9r '7, which is determined by the iUStitlltiOIlS and Public policies of an economy, and thereby

an 1 e I ‘ able to change.

is the collateral that the borrowers are willing to put up. We also make the following

additional assumption,

. . . . b 12k a
A 6 Collateralzzmg 23 costly- Collateral of value C; to the borrower gwes to the a

value of [3a, in the event the borrower defaults, where [j E (0,1).3

. t -
As in Hellwig [46], the a 3 stage sequential game is being played in the Credit marke 1
This game explicitly takes into account dynamic reactions of borrowers- The Opmna
loan contract can be Seen as a Solution to a 3 stage game,
Stage 1: Banks offer contracts (7" ’ C‘l' ex
Stage I]: Given the con tr acts the borrower i chooses p such that it maximize es he er we:
to
Pected revenue from the projed' net Of (a) repayment COSt (b) eflort costS- T ebOt
chooses the contract most attractive to him. He can choose only a single CO “ac" .
waved in

Stage III : The banks may accept or reject the loan applications they have ‘

Stage ll.
The Optimal contract (73,51); Vi = H, L is a set of contract Offers wh‘ ,
1C1: determine
the equilibrium in this 3 stage game. i.e. given (1%, (3,.) Vi = H, L
- Banks make zero profits on each contract and

_ N 0 bank has the incentive to offer a different loan contract than the one
Offered.
Since this game is sequential, the equilibrium considered will be the Sn};
ga e perfe t

01‘ Sequential equilibrium. The usua1 condition that each agent’s strategy be th
e hes st

 

response to the other agent 8 strategies is applied not only to the OVerall game but

BT11 6 assumption of the bank not being able to realize the fun value for the Collateral in C85
e of
(lefa‘Iu by the borrower, apart from having real world justifications is also crucial for the existence 1
. . ' ' 0

mini 1 ibrium in a risk neutral environment. See Clemenz [26].

52

to every decision node in the game tree, regardless of whether this node is reached in

equilibrium. We define Uij to be the utility to agent of type i applying for a loan contra”t

meant for type j.

2.3.1 Equilibrium When the banks can distinguish types

The
s.
As a benchmark, we solve first the (335%? when banks can distinguiSh between type

. e aware
banks are aware of the illegal payments that law: to be made by firms, Le they at

f
. . bsence O
of Whether the revenue of the firms 15 YF 0T YL. This is the solution 111 the a

rd
al haze»
adverse selection, when banks have t0 tackle only the post, contractual mot

_ they
after
problem. They are not aware of the effort that will be put in by the borrowers

get the contract. The problem for the entrepreneur is,

2
Man: U“; = PiKYi — Til " (1 - p-i)Ci — E; 3.1:.
Pisriaci

ill the Zero profit constraint for the banks: Rb E pm- + (1 “ POfiCi = 1.
my the seli selection constraints U,,- 2 U , j.

(iii) The Individual Rationality constraint U5,- > Z.

In the full-information case, the self-selection constraints do not bind.

he
I‘efOI.
bOrrower solves 6, the

- . .2
‘A’ICLIE’ ’ pi(yi _ ri) _ (1 — pi)ci _ p?
P2571461

S-t- Rb E Piri + (1 — PDBC, = 1

U?’. 2 Z

11

wllere 1); refers to the equilibrium Choice of effort by the entrepreneur and U?-

” TQfErS t0

the equilibrium level of utility, once the effort has been chosen. Before we give the f n
, u

53

 

 

information solution we see that the equilibrium effort level pg, given the interest rate 7'"

and collateral c,- has to satisfy the following conditions,

2.1)
P:=yi —7‘i+ci where 0<p:_<_1. Vi. (
.. _,_ (192')2 (22)
Uii :2 2 “ Ci = Z. Vi.
(1R) condition

(1) is the condition on the probabilities and (2) is the individual rationality

at the equilibrium

which determines the entry of the POtential entrepreneur. We see th

0t only on terms 0

80a

level of effort, or the probability Of sueCESS of the project is dependent n

- . , ‘n65 Y"
the contract, but also the CffiClCnCy 0f the social infrastructure which dctcflm

iuncl’lon’

P = 190/137"an The above formulation thus enables to get a simple probabillw

. mefit
\HIO“
where the probability of success (effort) is positively related to the business e

coha‘e‘r‘“
within which firms in the economy have to carry out. their business pro jectS,

hich they

which they are 35kt ‘0 DOSE and negatively related to the gross interest rate W

have to pay on the \oan. The full information solution is

Ci :1: w.

(23}

n- _._ l{}/i+(1+g)w—[(Yi+(1+B)w)2+4(/3w(1_Y—w)
- (2.4)

There are two things to note about this Full Information Solution:
(j) Though the maximization problem involves solving 3 unknowns giVen
a System of
2 equations - one of the unknowns, namely Ci, gets determined by the nature of t
he game
beillg played in the credit market. The banks would prefer to take maximum collate 1
ra

erII1 all agents (c,- = w). The reason being that the expected returns of the bank R
b

 

 

'ncreas ' - ' ' ‘
1 es With the collateral posted, VIZ. @5sz > 0, but g—f? lS ambiguous. 511106,

 

 

6
6R1) 3 .
_ p1 613' 2 5)
— i + z‘ — i — ' ( '
8r,- 1) (7" 56 l 67} where (973‘ < O
The effort that the entrepreneurs put in, which defines the success probability (P) 15
k5 would

negatively related to r, but positively related to the collateral, c. Thus the ban
require the borrowers to use all their wealth as COllater a1, and determine the interest rate
SUb‘jeCt to this‘ Loan applications With any other amount of collateral would be related
by the bank in Stage III and knoWing this’ in eQUilibrium no such application would be
forthcoming The 3011113011 ‘30 Ti in (4) is then a solution to a quadratic equation. “ 15
bank, given

obtained by plugging p: and Ci ; 'w in the the Zero profit condition of the
in (i) 4.

.. e deflmd
in) When the banks can SBPaIate Out the types , individual COntractS can

by the banks, such that corruption Can be weeded out to an extent As we See bd

firms who engage in corrupt practices are charged a higher interest rate mpared t0
, as Co

the, more honest firms.

67} 0
8K < => TF < TL

(2.6)

 

 

4We take the root with the lesser value. Both the lenders and the borr kn
OWEI‘S 0“?

garne, and if there exists more than one solution, the lowest among these (viz th 8 rules of the

o o S ~
the high” effort. p ) Will be compatible With pareto efficiency. OCzated With

 

2.3.2 Asymmetric information equilibrium in the Presence 0f

both adverse selection and moral hazard:

' l
The full information equilibrium does give us a clue into the equilibrium that we W111
, n
reach when the banks are unaware of the reanueS of the firms who apply for loans
considering the separating equilibrium as a SOlution, we see the follOWinga as
w ,
In the r—c Plane, totally differentiate U; = ((1%L2 ~ Ci) to get in the context of bOIIO
£12 a 22* — 1 (2.7)
dq T T < 0 d
Wwaf
(7) eStabHSheS that the indifference cur ves 0f the borrowers in the r-c Space are do t make
0
sloping, and satisfy the single crossing property. viz. the borrower who expects d m the
axe

unofficial paYments and get a lower return YL: has a lower MRS 35 Comp
. ~ - duC‘ilo
borrower who eXDeCtS his full Yleld from the pro J ect Yp. For a small re
- 1m 1 tome“
interest rates, the \atter is w mg to post more collatera as Compared to the
condition f th ist - . - . .
However, the other or e ex ence of a separating ethbnum IS that
the zero—profit Curves for the banks should be flatter than the indifferenc
”-"Ves of the
borrowers. We show that this condition cannot be satisfied and therefo
, e’ the tYpes
Cannot be sorted out by the banks. The contracts cannot be self selected

. and we have
tlle following, [Proof in the appendix].

T1 N art-existence of a separating eqUil'ibTium in the credit market

W’ll en collateral is also used to monltor the effort put 111 by heterogenous borrOWers after
a loan contract is accepted, collateral ceases to be a screening device to screen Out the
em cient borrowers from the inefficient. Technically T1 establishes that in this mode], the

56

 

zero-profit, curves of the banks are steeper than the indifference curves of the borrowers:
wlliCh violates the necessary condition for the existence of a separating eqUflib’ium'
What this means is that in spite of the the bank having an additional instrument in the
form of collateral, it still cannot have perfect control. It is not possible for the banks to

013
in this Case, can“

separate out the corrupt firms from the non-corrupt firms Collateral,

n behind this result is

be used as a self—selection device by the borrowers The intUltiO
oweI‘S put

that collateral is akin to “monitoring Costs” in this model. BY making the b0“
S Of both

in all their weaIth as collateral the banks are ensuring that the emreprene“I
T311 banks

types put in maximum efTOrt after the Contract is accepted. By taking collate

try to solve the moral hazard issues which arise in the credit market.
. . QfiSid‘eYEd
Thus, in this model where both adverse selection and incentive effects at
u . . “0‘, be
Simultaneously, exxs‘tence of private Signalling in a separating equilibrium
5 not, selve

possible. Collateral is used by banks to solve moral hazard problems _ it doe

as a screening device to screen Out the corrupt firms from th
e non-corrupt The Ollly

equlllbfium which exists in this economy is the complete pooling equil’b '
l r .1th
, Where only

one type of a bank contract is offered to everybody,

Pooling Equilibrium : The solution to this pooling equilibrium is gi
Veh b
5’

C211}.

T = it? +(1+ mu; —— [(Y + (1 + (3)112)? + 4(fiw(1 — Y ~ w) _ ml}
where Y E Yp — 7(Yp _ YL).

OIIQ e we have established that no separating equilibrium can exist the solution t l
’ O t 1e

57

 

 

pooling equilibrium follows the same logic as given in the Full information case (3) and

(4). However, since the banks are not aware of the actual yields that the entrepreneurs

get from the project, they have to offer the contract on the basis of the average Yield

. . , . ture
that the perceive in the economy, depending Upon the .7, or the 50031 infrastruc

d by the banks'

parameter. So, Y is the average productivity in the economy as perceive
. . . - he Full
The reason for taking mammum collateral (C = 10) remains the same as given in t

. . . . - Stage
information case, where Since loan applications can be rejected by the banks 111

III, therefore in equilibrium, no applications with c < 11) will be forthcoming

- n
5 . - hen Ewe
For ease Of comp utation WLOG assume 5 = 1. The Pooling equilibrium ‘5 t

by

\‘2 .8l

C5111).

(2.9)

r = Erna—ire 4 41
2 + 111—12}.

3", the average yield of the prOject in the economy, is defined as befor
. o.

2-4 Social Infrastructure and Credit Ration -

We now introduce another form of heterogeneity in the model We assu
. me t
L - ' T», . - . at "‘diVid‘
ld-ls differ lIl wealth (W). 1118 llCtel ogcnmty 15 ObSCIVablc and diflcrcnc
'08 in t}
10 WC'
dlth

of borrowers is public knowledge. We assume that the wealth is distrib t d
1381‘ a

s H < 1 is the necessary condition for 3 Of a separating equilibrium in a risk
neutral enviroiim,ent
Pr 1 13' ' ‘ ' - .
Q of T1 Slows that even when [3 E (. 9 1) a poolm ecui]
of g l ibrium pareto dominates. Therefore We

Call 355111118 [l = 1, WLOG.

 

 

 

 

 

 

 

C'd'f F(w) We retain the a‘SSUInptiOII Of all wealth being used as collatera/in borrowing-
Lendel‘s can now 013551” borrow” as per risk, on the basis of an Observable indicator,
the wealth they have to pOSt as collateral. Given that collateral is positively associated
with effort. (and negatively associated With risk), the banks Classify the poor borrowers to
he riskier than the rich borrowcl‘S- This Classification determines the interest rates that
will be Charged to each class. We also get the familiar result[10] that the interest rates
in this economy are negatively related to the wealth that can be posted as collateral.
There exists credit rationing in this economy, such that all borrowers below the critical

wealth group 10* do not have access to credit. We Car) Show th f mowing [proof in the
e o v

appendix]:
P1 In a pooling equilibrium in the credit market, the 3 Qcial infrasthWTe

of the economy determines the critical 212*, which defines the extent of credit

rationing in this economy.

The intuition is straight forward - given the Social infrastructure p ararn t
8 er 7 of
o n e an
economy, collateral is used then in monitoring the effort exerted by the
. . n9111‘s
1n such an economy. Entrepreneurs posting more collateral are seen to b

e Safer I'iSkS
)

i Frespective of their actual returns from the project and are charged lower in t
Brest 1'8,
1563.

(7) also determines the critical wealth group wt (calculation in the appendix)
W

which no lending will be done by the banks. All borrowers having Wealth less t}
K lax] wt

irrespective of their actual yield from the project, Will be rationed fr Om the credit
market

Tl‘lis will occur in spite of the fact that lending is SOCially Optima], It is Perceived b
Y the

})E\1*1k8 that lending to this group does not make the banks brealoeVeIl at any interest
rate.

59

 

 

 

 

10* is a function of the sociaI lllfl‘astructme index of the econOmy 7, Higher the value

of this parameter, greater- will be the proportion of the population that will be barred

from the credit market. A high ’7 means that a greater proportion of firms have to
indulge in bribery and other unofficial payments - this indicates the quality of the social

. ref -
infrastructure In the economy. The ore, Elven a distribution at wealth in the economy,

the extent, of credit rationing is greater When the Social infrastructure oi the economy is
Poor (7 is high). The banks will then play safe by lending Only to the wealthier groups

having Sufficient collateral. This is ShOWn in the diagram below Wh th . t rest rates
ere e m e
in the economy are negatively related to the amount of Wealth DOSt d llateral given
e as CO 1

the economy’s social infrastructure indicator present in 7, W

ith a decrease in 7’ and
thereby an increase in )7, there iS a Shift Of this curve to the left F . en distribum“
- or a glV

'Y decreasing

/

Figure 5: Infrastructure and Credit Rationi
n

 

of wealth, lower the 50031 infra‘Sthture Parameter 7, both the cost of funds 7' and w',

the critical wealth level below Which borrowers are barred from borrowing, is reduced-

The extent of credit rationing 0f borrowers below a 112* thus depends upon the social

inf-1'35tr ucture indicator of the economy,

2.4.1 Effect of a decrease in the Parameter ’7

From the section above it is now easy ‘50 COHCIUde about
3 the (3 ect. . . n
, fl , s of policy (lI'IVC
initiatives which reduce ’y, the proportion of firms who have to t . l
resor to unofHCIa Pa3‘

ments and corruption, viz. the size of the ‘shadow’ or unoffici
a] economy.

e see that a,

with a decrease in the parameter '7’ the COSt Of borrowing in is reduced
the economy .

Therefore, this result suggests that economies ranking low on t 1’1 . dex would
Q corruptiOn ‘“

also be able to obtain funds at cheaper rates than those econQ

1'68 which are plagued by

corruption and rent seeking act1v1t1es.

(b) There is also an “entry” effect, related to the extent of Cr

am With the decrease in 7, the cut off 111' is reduced, 3'
.Y- "Ice

rationing is a possible outcome in many models of asymmetric .
"If .

b - ' ' ' h f corr t’ ' - Canal: ~ [V6119

1111:), 15 made severe in t e presence 0 up Ion, meffiClencieS . 1012

111 j ’ £71.

”In 18

. , , . '. , I .00. .
IQWS: and all those factors Wthl'l hamper the firms from Con duCting t $173th 5191‘

6261‘
. O
elr I] 0

1"

T1 1erei0fe policy initiatives to reduce 7 will ameliorate the degree
Of ere . eSseS

SL1 Q11 economies.

These results are important because they suggest that as as),
me

61

 

 

 

 

mOdelS about the credit market are made more realistic to include a range of information
asymmetries the Scope for pI‘iVate signalling is reduced. Public Signals in the form Of
laws and regulations, infrastructure, and the quality of bureaucratic services will then be

increasingly used to determine the extent 0f investment in an economy. The relevance

of such public signals lies in the fact that in an imperfectly informed worm, they expand

the information set that is available to the economy.

2.5 Conclusion

The aim of this paper was to prov1de one among the many possible explanations to

for investrnent
common query: What are the reasons levels bfiing higher in a country

. . a country like Bangladesh? I . -
llke Smgapore as compared to the tried to explain this

. . t t f as mmetric information in the credit .
Within the con ex 0 Y arket, It iS fairly clear

that as asymmetric models are made more complex to inclu

8 both .
and moral hazard, the possibility a pooling equilibrium and t1) adVerse SBiBCthIl

e
reby Cred ‘

. - h ’
equilibrium increases. In thlS paper I also S Ow that m the rose t [atlbnin .
“Ce 01“ gm

layered asymmetric information structure, private signals can

005

get; 0 b
I"
use a [DU/t1;

Will 1,,» a greater reliance on public Signals. An effimcnt legal

a .
nd Judi"
smoothly functioning bureaucracy, good quality Of Public serviceS “1&1

. a 531le

1‘

‘ , . 1k “irregular payments” t . . ”Jen
{111115 are not compelled to ma e 0 get the” t

c . - - ' heir smooth functioni ' , ’ fir
91.1}31'10’ ludicators which underlle t 11g. It IS fdctor e

S like tilese
arQ increasingly defining the lending and investment decisions of today 8’ hlch
. . lllgapOre .1
s

' O I V able
to I) ublicly signal its intentions to investors. The same cannot be S

end about. Bangladegl
‘ i.

62

 

 

Z-A Appendix:
1. Proof of T1:

The slope of the indifference curves in the r — c plane is given by totally differentiating

Uii a which is given as follows,

Uii 2 (Y; ‘ Ti + C-)2
N _

2 Ci
515 2 W
dci(borrowe1"3) (K ‘ 7}- + Ci) ‘< 0.

Since (3”, _ 7-,. + Ci), we know is the equilibrium probability f
0 success of the pr0j eCt
denoted as p“, and p" < l-

The isorevenue curves of the banks is given by

 

 

 

Rb : p‘ri + (1 — Pvflc,
dr 2e
EC—(banks) : %
BR 613*
7: = ac (7" “ ‘3‘”) + (1 ‘ 10W? > o
* V
aRb 6p * . ,6 e 0
ET :2 01‘ (Ti _ laci) +p % 0 SZnCe ap‘ ( 7 1).
am) ' i < 0
‘5; > 0 sz (Ti — fiC-i) < p” ,
iff 1 >fi>""” >0

 

C1:

Which puts very stringent conditions on the returns from the
i, .
\qz‘

2(ri __ Ci) < K < (2ft ~ Ci)

Q iVen the above, there is an additional condition on 5, (2.10)

, 'r~— “
1>r3>' p Efi’

Ci

 

(2.11)
63

 

 

' ' 'ons on Y'-
Given the two concht1 ' and 5' above, we will have the isorevenue curve of the

banks negative in the H: Space-

dr‘
dc (ban/CS)

=—W}l

2.12
p, g (a g M < )

Given that both the indifference curves 0f borrowers and the isorevenue curves of the

banks are negative in the r-C Space’ the necessary condition for the existence 0i a sepa—

rating equilibrium is that

dr
dc

/""/
(banks) dc (boy‘rmers) (2.13)

 

 

i.e. the indifference curves of the bOITOWGTS ShOUId be Steeper 1 th the
1n t 1e r—c space an

isorevenue curves of the bankS-

 

 

 

 

 

d7"! = 1‘ p"
dC (borrowers) p‘
E: = fiflp—P)+(7‘i~,ec
dc (banks) - (r-— —,6N

We new show that 2.13 above cannot be satisfied. Taking th
terms, we see that p" — (Ti " BCi) < 19.7 Therefore, the I1
existence of a separating equilibrium depends upon the value
0

the 12 (11,870,) [01‘ tile
Q

the termS- Which is, (11”

3(1—P’)+7‘i—5Ci < 1‘19

[3 < W
(l‘pt)‘Ci

VVQ know (1 _ p“) _ 7.1, < 0’ since r.- is the gross interest rate and theref0
Ore

> 1
t Elkil’lg the denominator of the RHS, we now have to consider two p herefOre
OSSlbiliti
es, (1

mu

64

 

 

Ci 2 0-

if (1_p‘)—q>0 2:} fl<0
if (1_p*)—q<0 ==> fi>1 (since l1_pi,n\>\1-pi—cil)
Which is a contradiction to our assumption A 3- Therefore, even ii the isorevenue curves
of the bank turn out to be negative in the r—c Space, a separating equihbtium cannot
exist as they turn out to be steeper than the

indi
fl’erence curVes of the borrowers I
II. Proof of P1 3

22:1 1

_N
8w (Y+4w —4):1>'

(2.14)
In order to prove that g:- < 0, we have to show the following ,
0 < Y + 4w — 4 < 1.
(2.15)
Note that from the definition Of T given in (7): we have
Y+4w—4 E l7~ 2(r— 20)
Y n 20" ..... w) > 0, which follows from the proof of T1. And ,t
of A OHOWS f
p‘ thatO<Y+4w—4 < 1. So, we see that, fort)
6,56,
_8: < 0 Mamba
0w '
i Veil a distribution of wealth F (W), there is no solution to (7
17 + 421) - 4 g 0.

TI) i 5 defines the out off w“ for which there is no solution to (7)
7 and ‘ - ,
Y 1t ‘3 Elven by
w“ E 1— Exp _ 79f ~ YL)

<2.
65 17)

_Im

 

 

What this means is that that below a COUateral level of 10', banks are not; willing to

lend Since they cannot break even on such contracts. Therefore, all agents with wealth

° (21”.:
E [0, wt] Will not be able to borrow even If they Want to. AS we see, 0w > 0, Therefore

the critical point of credit rationing gets lower With a reduction in the parameter 1 I

66

 

 

Chapter 3

Financial Constraints -
ture and
InfraStruc

0')
What do firms have to say.

3.1 Introduction

I tron I - ' 's an
. . 11 and p Oductwrty 1
5 11k between infrastructure p OVISIQ r
' 1i 5
t s g
hat there an

' linkages between in
f t However, the prec1se
estabhshed ac .

rod“ . . are

x Y . ts.
d ctivit 0f infiaSthture In CS 1116“
u
to CSfiifia ing the pro

The

emphas .
ro econometric studies (using both time series :nd Cro
ac - .
these m ‘ at “credible” measures of the impact 0 lnfi.
On amvmg 1
Productivity and growth.

eConomic
studies in the US and other developed e00
time-series
Most

001221.68
f t m on infrastructure deVCIOpment This ‘

rates 0 re u

Startli 11eg high

I ggest
4 . Man StUdies re
many cases (Gramlich, 199 ) y
' tion 1n
Overéstlma

u
rtp bliC Seetor
bstantially greater impact on private Sector Outpu
su
t as having

t t ivate
r uld bet two-
6c t. he reasons for this overestimation co
t stmen ‘l'
S Q ““16

fold: (a) Missing
1 pment Report, 1994 does a succinct summary of the various eStimates
1 1m
The BOx

67

 

factor explaining trends: There COUld be a common factor that causes changes 1.” b0 t1;
infrastructure and the output that needs to be included. Gramlich in his review essay cited
abOVe suggests that gasoline Prices which led to both, a reduced demand for
traCtors/trucks and thereby a reduced demand for highways and to a lower output in the
1970’s should be factored into this link. (b) Reverse Causality: Another important
CO“Gem has been endogeneity and the direction of causality between infrastructure and
output. While infrastructure may affect productivity and output, economic growth can
also shape the demand/supply of infrastructure services which could lead to the
OVBI'estimation of the returns to infrastructure. While common trend is a potent problem
in the studies which use time-series data, the cross section studies do arrive at more
SenSible measures ranging from an implied rate of return equal to the rate of return on
Pfi" Elte investment (on the higher side) to zero (on the lower side). But cross section data
3150 i s not immune from the problem of reverse causality. In addition, in the cross section
Studi es heterogeneity is another problem. There could be an overstating of infrastructure
imp acts by confounding intrinsic state/nation productivity differences with the variation
in infrastructure capital. Taking state level data for example, because prosperous states
W0111d tend to spend more on public capital, there will be a positive correlation between
State specific effects and public sector capital (Holtz-Eaken, 1994). However neither the
time series not the cross sectional studies explain the mechanisms through which
infrastructure affects growth. Unless such “micro” links between infrastructure and
grOWth are uncovered — it will be difficult to understand the complex aggregate
1'61 ationship. Thus these results suggest two kinds of agendas for future research in this

area; (a) detangling the endogeneity and heterogeneity issues econometrically using more

68

 

 

disaggregate data and (b) making the microeconomic linkages between pro‘mbn of
in frastrucmre and the nature of the production process more precise. It will b e (new to
know how other variables that affCCt growth work through infrastructure.

This paper tries to fill the gap as regards agenda (b) by using disaggregated nation

Wide firm level data based on the World Bank Business Environment Survey (WBES,
2000). It tries to show how infrastructure can crucially affect the ease with which firms
can obtain finds from the capital market. The theoretical underpinning of this link is
baSed on the analysis of public inputs in the context of asymmetric information, where
infraStructure provision is seen to reduce the heterogeneity among firms. This reduction
in heterogeneity has a crucial impact on the capital market, which is besieged by
impfil‘fect information. Infrastructure is seen there to reduce the costs of asymmetric
infoKination and improve the equilibrium in the capital market.

Regarding empirical studies on these infrastructure-financial sector linkages, there
hav e been some cross-country studies bringing out the link between infrastructure and the
fitlancial sector. Cross-national studies considering the impact of infi'astructure on
differences in the FDI flows (Globerman / Shapiro, 2002) find that FDI inflows respond

Positively to good governance infrastructure and human capital. In a more detailed study
regarding physical infrastructure and capital flows, Mathias Hoffman (2003) shows that
di fferences in information and transport technology is able to explain cross country
Variations in FDI and debt positions of countries . Sectoral studies focusing on rural
in fiastructure’s impact on the local economy in certain developing economies have
re\Iealed more about the infrastructure — financial sector linkages. Studying data over

time from 85 districts in 13 Indian states, researchers found that improved

69

communications (through roads) lowered the banks’ COStS 0f ‘10ng busineSS Banks
expanded lending to farmers and thereby increased farm output (Binswangef, et 31’ 1994).
This study differs from the above in several ways. It uses firm-level data to

Capture the link between infrastructure provision and financial constraints. The data is
qualitative, based on the responses given by firms to questions regarding the difficulties
faCBd by them with respect to infrastructure provision and obtaining finance. Using this
Slime), data from some 10,000 firms spread across 80 countries (The World Bank
usiness Environment Survey, 2000) this study tries to capture the nexus between these
infi‘aStI'uctural and financial constraints facing firms. Using both Ordered logit and
Ordered Probit estimates, it concludes that taking care of all region specific and firm
Speci fic differences, firms facing high infrastructural constraints are most likely to face
high financial constraints as well. More specifically, this link is seen to be stronger in the

case

of (i) firms in low income, developing countries as compared to high income

dev eloped countries (ii) smaller sized firms as compared to the larger firms.

3.2 Data and Methodology

The World Business Environment Survey (WBES 2000) was administered by the World
Bank in roughly a parallel fashion to enterprises in 80 countries and one territory
thl‘oughout the world, as basis for making regional comparisons of investment climate
and the business environment conditions. The World Business Environment Survey
(WBES 2000) is a survey of over 10,000 firms in 80 countries that examines a wide

1"ange of interactions between firms and the state. Based on face-to-face interviews with

70

firm managers and owners in late 1999 and early 2000, the WBES 8mm
measurements in such areas as cOrI'Uption, judiciary, lobbying, investment 01inl ate an d the
quality of the business enviroflInent. This survey thus tried to capture the my
perceptions of key constraints in the business environment. Among the constraints faced
by the firms were those relating to (a) Finance and (b) Quality of public services. Finance
was the second leading constraint for most firms. At least 50% of the firms in all
eVeloping regions cited financing as a serious constraint, only 40% of the firms in
015C D countries found it to be so. Finance as a constraint was more important to the
SIhall and medium enterprises in the survey than to the large enterprises. Another key
dim€>erlsion of the business environment was the quality of public services. WBES
exPlQred both the overall efficiency of the government in delivering services and the
quality of individual services. Nearly two thirds of the firms in Central Europe, Latin
mefica and CIS countries and nearly 60% of the firms in South Asia report that
30" ernment is inefficient in delivering services.

To get an idea of this nexus, table 1 shows the correlation between financial
cgtlstraints and infrastructure constraints. Taking the subset of private, domestic firms
(State owned firms and foreign firms were dropped out) it is seen that while 35% of firms
fElcing no infrastructural constraints reported they had no financial constraints too, while
OIlly 7% of the firms facing major infrastructural constraints reported that they no
fillancial constraints. A test of the null hypothesis using Kendall’s tau-b that financial

Constraints and infrastructural constraints are independent was rejected.2 The Kendall’s

taill—b was more significant when this test was done disaggregating the data by firm size.

\

2 _Kendall's tau-b = 0.2441; Kendall's score = 3654854; SE of score = 157568.200 (corrected for
t1€28); Test of Ho: gcf and infr are independent; Prob > |z| = 0.0000 (continuity corrected)

71

Small sized firms showed a stronger correlation betWeen infrastructure and fame. I
1a

Constraints as compared to the medium sized or larger firms.

Table 1; Correlation between Financial constraints and Infrastructure constraints

general |
constraint - | general constraint - infrastructure
finance | no obstac minor obs moderate major obs l
“~~__ ____________ + ____________________________________________ + _______
no obstacle | 671 218 146 77 |
minor obstacle | 319 401 254 109 |
moderate obstacle | 414 579 524 252 |
major obstacle | 530 619 650 633 |
“-~ __ ____________ + ____________________________________________ + .......
Total | 1934 1817 1574 1081 |

Pearson chi2(9) = 799.3153 Pr = 0.000

l3j~1<elihood-ratio chiZ(9) = 750.6341 Pr = 0.000
Cramer's V = 0.2039

gamma = 0.3313 ASE = 0.014

Kendall's tau-b = 0.2441 ASE = 0.010

3.3 Regression Estimates

For getting the regression estimates, the methodology followed was thus: It is being
hypothesized that the financial contracts obtained by the firms is determined by firm
specific factors - its size, debt, sales and asset position, its future prospects; region
specific factors, namely the political situation in the region the firm is situated, the
monetary and fiscal policies of the region, and the infrastructure situation as

perceived by the firms. So the financial constraints faced by firms was regressed on

72

(a) infrastructural constraints (b) firm specific factors whiCh determing f. . I
mancza

contracts, specifically _. value of sales, fixed assets, debt (all taking 1 o g S) size

Characteristics, firms’ perceptions about past and future sales, investments and debt
Values and (c) region specific constraints which could influence financial contracts,
Viz. the ranking of the nation regarding income, the fiscal and monetary policies of
the country, the political instability of the region, the quality of central government,
and the corruption levels perceived by firms. (A detailed definition and description of
the variables used is given in App. A). Tables 2 onward show both the ordered probit
and ordered logit results. We are interested in the relation between the dependent
Variable being the financial constraints facing the firms (gcf) and the infrastructural

chstraints facing them (infr_d).

Tib 1e 2: Ordered probit estimates on the General Financial Constraints (gcfl facing
knv ate/domestic firms QV__BES,2002).

Eumer of obs = 3367; LR chi2(24) = 642.34,- Prob > chi2 = 0.0000;
0% likelihood = -4047.7911,~ pseudo R2 = 0.0735

 

 

gcf | Coef. Std. Err. z P>|z| [95% Conf. Interval]

‘ ‘_ ————————— + —————————————————————————————————————————————————————————
cQ‘untryl [.4318122 .08244 5.24 0.000 .2702328 .5933916
QQ‘untryz I .353737 .0630527 5.61 0.000 .230156 .4773181
qgov_d [.1161673 .0407323 2.85 0.004 .0363335 .1960011
ngi_d I .06682 .0475951 1.40 0.160 -.0264647 .1601047
inf1_d I .1088632 .0474302 2.30 0.022 .0159017 .2018248
txreg_d .3476044 .0525174 6.62 0.000 .2446722 .4505365
gcorr_d| .2406343 .0434523 5.540 0.000 .1554694 .325799
tgdnLregdl .1709161 .0447298 3.82 0.000 .0832473 .2585849
81261 .2976406 .0696898 4.27 0.000 .1610512 .43423
81262 |.2424178 .0669084 3.62 0.000 .1112797 .3735559
lnvsal I .0030652 .0108683 0.28 0 778 - .0182363 .0243668
lnvfas |.0135233 .0111458 1.21 0.225 -.0083221 .0353688
lnvdebt I-.024300 .0078258 -3.11 0.002 -.0396385 -.008962
fn_re |-.000853 .0005122 -1.67 0.096 -.001857 .0001506
sal_d [- .037877 .0528281 -0.72 0.473 -.1414185 .0656638
Sa1f_d |-.021899 .0546672 -0.40 0 689 -.1290451 .0852465
inv_d |-.124489 .045451 -2.74 0.006 -.2135713 -.0354066
ihvf_d [.0746657 .0485123 1.54 0.124 -.0204167 .169748
1ab_d |.0570087 .044256 1.29 o 198 —.0297315 .1437488

 

labf_d \ .0132834
debt_a \ .2412359
debt} \ .1342956
\ -.064562
| .3772739

[“90 obstacle |
nwlnor obstac |
mofierate obs |

aJO]: obstac I

Pr(
Pr(_cut1<xb+u<_cut2)
Pr(_cut2<xb+u<_cut3)
Pr(_cut3<xb+u)

.0445556 o
.0469151 a
.0472459 2.
.0432783 -1
.0426074 8.
.2318569 .1232675
.8395603 .1236247
1.66778 .1251726
Probability
xb+u<_cut1)

.30 0.766
14 0.000
84 0.004
.49 0.136
85 0.000

— . 0740635
.1492839
.0416955
—.149386

.293765

‘.1006303
~£flUU8ZQ

~25$3958

.020262

.4607828

Table 2.1: Sample Statistics of the Financial Constraints facing firms : Probit Model

Percentiles
1% .016805
5% .0264117
10% .0363703
25% .0611312
50% .1117227
:5 5% .1996599
0 % .3180288
95 3: .3877719
99 95 .541106
Percentiles
‘3t% .0478262
5 % .0655686
1-C>% .0812925
2 5% .1130827
5 0% .1593029
7 5% .2067256
9 0% .2321494
95% .2374023
9 9% .2387057
Percentiles
3.% .1613007
5% .2018408

Pr(gcf==1)

Smallest
.0085028
.0111069
.0113076
.0114367

Largest
.6491151
.6966723
.6983643
.7171385

Pr(gcf==2)

Smallest
.0291266
.0354702
.0359377
.0362371

Largest
.2387582
.2387586
.2387588
.2387591

Pr(gcf==3)

Smallest
.0963894
.1045309

74

Obs

Sum of Wgt.

Std. Dev.

Variance
Skewness
Kurtosis

Obs
Sum of Wgt.

Mean
Std. Dev.

Variance
Skewness
Kurtosis

3382
3382

.1486163
.1181638

.0139627

1.44125
5.047452

3382
3382

.1577637
.0550815

.003034

-.1749869

1.907029

10% .2274336 '105271 Obs 3382

2 5% .2650309 . 1264649 Sum of Wgt. 3382
5 0% .2979814 Mean .2847372
Largest Std. Dev. .0388117
'7 5% .3158797 .3212062
590% .320386 33212062 variance .0015064
95% .3209755 .3212062 Skewness -1.374217
99% .3211944 .3212062 Kurtosis 4.538319
Pr(gcf==4)
Percentiles Smallest
1% .0618846 .0222005
1 5% .1249089 .0252547
250% .1678479 .0255418 Obs 3382
% .2765617 .0344667 Sum of Wgt. 3382
50% .4135177 Kean .4088827
7 Largest Std. Dev. .1745411
5% .5435668 .7994395
90% .6400273 .8006518 Variance .0304646
95% .6916173 .8025514 Skewness -.0359967
99% .7545248 .8291127 Kurtosis 2.169749

Tap le 3: Ordered Lo it Estimates on the General Financial Constraints c facin
\Pr1Vate/domestic firms (V1883, 2002)

‘Nutrfflber’of obs = 3367; LR chi2(24) = 637.53; Prob > chi2 = 0.0000;
Log likelihood = -4050.1943; Pseudo R2 = 0.0730

 

gcf | Coef. Std. Err. z P>|z| [95% Conf. Interval]

\ - __________ + _________________________________________________________
‘=‘=>untry1 I.7526405 .1378575 5.46 0.000 .4824448 1.022836
5391.111th I.5926867 .1040164 5.70 0.000 .3888182 .7965552
qgov_d I.2097505 .0682455 3.07 0.002 .0759917 .3435092
gcpi_d I.1136429 .0797983 1.42 0.154 -.0427588 .2700447
infl_d I.1775468 .0794326 2.24 0.025 .0218617 .3332319
t:xreg_d I.5747604 .0882174 6.52 0.000 .4018574 .7476633
Qcorr_d I.3910674 .0731885 5.34 0.000 .2476206 .5345141
t1!:thn_regd|.2767492 .0751361 3.68 0.000 .1294851 .4240132
sizel I.4979607 .1162416 4.28 0.000 .2701315 .72579
81262 .4163502 .1116201 3.73 0.000 .1975788 .6351216
lnvsal I .0029228 .0182491 0.16 0.873 -.0328449 .0386904
lnvfas I.0232014 .0187301 1.24 0.215 -.0135089 .0599117
JLlfrvdebt |-.038626 .0132012 -2.93 0.003 -.0645 -.012752
fn_re I-.001455 .0008634 -1.69 0.092 -.0031471 .0002372
sal_d I-.038096 .0890218 -0.43 0.669 -.2125754 .1363837
sa1f_d I-.048070 .0918659 -0.52 0.601 -.2281242 .1319836
inv_d I-.231266 .0765298 -3.02 0.003 -.3812618 -.0812706
invf_d I .110892 .0815559 1.36 0 174 -.0489547 .2707386
lab_d I.08773'75 .0740175 1.19 0.236 -.0573341 .2328091

75

 

Liz"

labf_d \.0292579 '0745322 0 39 0.695 — .11687'75 .1753932
debt_d I.4010898 ° 07 4 6 5.14 0.000 3432204 ~5539592
debf_d 1-2233626 $785 ‘4 2.84 0.004 $594183 .37730
1n£r_d I.6261608 .0717253 8.73 0.000 ~435531 '7 33739
_________ +____________---_;1 ‘___-__-_____________---__ ‘ ---_ - 567393
cum I .3430544 05876 (Ancillary ------ ~
cut2 I 1.396923 2066047 parawneters)
cut3 I 2.765965 2106662
gcf I Probability "BL-8;};a -\ \ ----------
............. +___-_-----—--—— - - - ~ ~~------__-__ -____
no obstacle | Pr( xb+11<_CUt1) 0 1491
minor Obstac | Pr(_Cut1<Xb+U<—°Ut2) 0 1571
moderate obs | Pr(_Cut2<Xb+U<—CUt3) 0.2839
major Obstac | Pr(_cut3<Xb+U) 0.4099

Table 3.1 : Sample Statistics of the Financial Con straints facing firms : Log.___._———————I'lt MN!“

1%

5%
10%
25%

50%

75%
90%
95%
99%

1%
5%
10%
25%

50%

75%
90%
95%

3382
3382

.1489362
.1114103

. 0124123

1 . 590841
5 - 713092

3332
3332

.1584705

.0621099

. 0038576
. 038 8674

1.81254

Pr(gcf==1)
Percentiles Smallest
. 0270871 .0172355
.0367092 .0208478
.0467015 .0213558 Obs
.0695522 .0214369 Sum of Wgt .
.1132746 Mean
Largest Std - Dev .
. 1912138 .6548852
. 3057353 .6909733 Variance
. 3768308 .7020864 Skewness
. 5394117 .7193993 Kurtosis
Pr(gcf==2) -__
Percentiles Smallest
.0468745 .0306657
.0618392 .0367163
.0765187 .03755'7 Obs t
.1070239 .0376911 3‘3““ Of W9 '
.1549103 mean
Largest: Std- Dev.
.2124117 .2575355 .
.2475532 . 2575355 Varlance
.2554115 . 2575355 Skewness
.2574555 . 2575357 Kurtosrs

99%

—-_
.— -—
'-—

Percentiles;
.1472111
.1865354

SmalleSt
.0862648
. 0925969

76

10% .2155495 - 0:3;304 Obs 3332
25% .2589253 - 1 528 Sum of Wgt. 3332
50% .2990818 Mean 3393154
Largest Std 045
. Dev. 96
75% .322472 .3294934 39
90% .3232934 .3294937 Variance 002 11
95% ,3292053 .3294939 skewness 1 200 2'7
99% .329472 .3294939 . 647
Kurt-.0318 3 850
178
1317(ng ==4)
Percentiles Smallest ----------------- \ \~-_
1% .0703788 03342 69
5% .127874 0362 58
2150: .1675938 .03813 99 Obs
.2727394 .04463 85 3382
Sum Of Wgt. 3382
50% .4097037 Mean .408273
7 % Largest Std - Dev. .1751218
5 .5425661 .8018758
90% . 6441087 .8024908 Variance .0310189
95% . 6993946 .8063595 Skewness .0231392
99% . 7610283 .8348624 Kurtosis 2.120622

ct tha‘ the

Tables 2 and 3 indicate that both logit and probit models predi

infrastructural dummy (infr_d) is positively significant in determining the credit

difficulties faced by firms. The estimate on infr_d being Positive indicates that “other

things being equal”, those firms who faced major infrastrucmral difficulties had a higher

probability of having mOderate/major financial constraints and a lower probability of

ms having

' n ' ' ‘ com ared to £1 -
havrng o/mmor financral constramts, P 110 or minor

infrastructural difficulties.

sc ' . 10” £30 .
The two models also suggest that among the region specif tors determining
_ . . ntri
credit constraints: Both low income and mlddle income co“ es had a greater
to high income countries.

PFObabilitY 0f having major financial constraints, as compared

- . . - ' on to middle

(The coefficrent was higher in the case of low income C0untr165 1“ compar‘s
- . . - - ountries had

inCOme comm“: indicating that Cetens Panbus, finns 1n the lOW Income c

77

 

the highest probability 0f having major financial constraints). Other region Sp eci {’1 c
facmrs Wm“ “11’“ 0‘1”" be Significant are the dummy for quality of central govemm
(qgov_d), the inflation dummy(infl__d) the durum}, for taxes and rcgul ations i ent
(txreg_d), the dummy for corruption (goon-4m and the dummy for tax admi “ a country
regulations (tadm_regd). Other things being equal, those firms which f mstration and
governance, high levels of inflation, hi gh Comption, and constricting :‘Cedl poor Quality
administr ' - - ' , ax 3W5 and tax
and a low:l::o:::iiit:l::a:ir::::::i:o:ffi::m'g mOderate/major financjal coma-aim ts,
Clal constraints. The dummy for Pounce”

msmbility (ngi_d) turned out to be insignificant in both the models.
Regarding the “firm specific” factors, the Size of the firm (sizel and size2.) and all

the debt variables, log of the value of present debt and the firms’ perceptions of the 13"“;t

and future debt levels turn out to be significant (lnvdebt, debt_d and debf d). Sma“ and

medium firms have a greater probability of having moderate/m aj or financial constraints

as compared to large firms. The finns’ existing debt position (lnvdebt) is negatively

significant in both models. Those firms who had higher debt values had a lower
probability of having major/moderate financiat constraints as Compared to firms who had

lower debt. Also those firms who had increase:d debt in the past 3 years and eXpect d
e an

increase in their debt levels over the 116m 3 years were more likely to h
ave

moderate/ major financial Constraints. This suggeStS that firms whose exisfing debt values
were low, and Who were expecting to borrow in the future are the Ones who are most
likely to say that financial difficulties are greater. The dummy for increased inveStment in
the past 3 Years (inv\d) is negatively significant in both modclS, meaning that those firms
whose tIlVC’Stment levels increased in the past 3 years had a lower probability 0f having

78

major/moderate financial constraints, as compared to firms whose investment levels did
not increase.

A display at the bottom of tables 2 and 3 show how the probabilities for the
categories were computed from the fitted equation. Notwithstanding the differences in the
coefficients between the logit and the probit models, the predicted probabilities are
similar, so in this case it does not matter which model is being used.

Table 4: Overall Probability 0f haVing financial constraints: Probit Model

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

CALCULATE? 'AS CALCULATED FROM
Mean of. individual Mean of Determining
Probability of PrObabllltieS (p) Variables (q)
having
No financial -1486163 .1209111
constraints 4______,
Minor fin. .1577637 .1658946
constraints
Moderate fin. .2847372 .3178739
constraints
Major fin. .4088827 .3953204
constraints

 

 

 

 

 

 

 

 

Table 5: Overall Probability of having financial constraints: Logit Model

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

CALCULATED AS CALCULATED FROM
Mean cm individual Mean of Determining
probability of Probabilities (p) Variables (q)
having
No financial .1489362 .1212659
constraints
Minor fin. .1584705 .1623435
constraints
Moderate fin. .2843154 ,3252247
constraints _____________7
Major fin. .408278 .3911659
constraints 4__f

 

 

 

 

 

 

 

 

 

 

79

Tables 4 and 5 show the sample statistics of the financial constraints facing the firms,
generated form the individual probabilities of the firms, using both probit and logit
estimates. The mean probabilities (p) in the case of the logit and probit models are shown
in the column 1 of tables 4 and 5 (Mean of individual probabilities, p). They are similar
and after rounding Off, and are equal to 15%. 16%, 28% and 41% respectively for the 4
levels of financial constraints. These estimates can be compared to the actual sample
proportions shown by the data. The sample proportions of the firms having
no/minor/moderate/major financial constraints are 17.06%, 16.87%, 27.83%. and
38.24%. The mean probabilities of both the logit and the probit models are thus close to
the sample proportions.
There is an alternative way of calculating the mean probabilities and this leads to
a different outcome from those set above. The intuition behind this method is
constructing a “straw firm” which has the mean value of all the determining variables,
and embodies the average attributes of the sample. This method by-passes the individual
probabilities, and calculates the probabilities directly as the probability faced by this
“straw firm”. These are given for both probit and logit in column 2 of tables 4 and 5
(Mean of determining variables, q).

Two features of this table are significant. First, for any two ways of computing
marginal effects there was hardly any difference between probit and logit probabilities.
Second, for any one model there was considerable difference between the probabilities
calculated in the two different ways. The probability of having no financial constraints
and major financial constraints is higher when the probabilities are computed as the mean

of the individual probabilities (P) than when they were computed from the average

80

characteristics of the firms (q). This is not surprising. Both these rankings are the result of
possessing “extreme” values of the determining variables. This influence of the extreme
values is dampened when the individual values are set equal to the sample averages. On

the other hand, the extreme values are allowed full play when the individual values are
used in probability calculations.

3.4 The Infrastructure - Finance link
The effect of the dummy variable (infr_d) Should be analyzed by comparing the
probabilities that result when the dummy variable takes one value with the probabilities

that are the consequence of it taking the other Value, the values of the other variables

remaining unchanged between the two comparisons,

Table 6: The effect of Infrastructure on the robabili rankin s of financial

constraints: Probit Model

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

CALCULATED as: CALCULATED as ;
Mean of individual Mean of Deterndning
Marginal Effects. Variables,
Probabilities Infr_d = 1 Infr_d.= 0 Infr_d = 1 ‘Infr_d = 0
(gc f )
No financial .0994008 .172585 .0796417 .1514553
constraint
Minor fin. .1317345 .1749444 .1322767 .1848717
constraint
Moderate fin. .2768532 .2954325 .2994248 .3211955
constraint .
Major fin. .4920115 .3570382 .4886568 .3424775
constraint {

 

 

 

 

 

81

 

Table 7: The effect of Infrastructure on the robabilit rankin s of the financial
constraints : Logit Model

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

CALCULATED as: CALCULATED as:
Mean 0f indiVidual Mean of Determining
Marginal Effects. Variables.
Probabilities Infr__d = l Infr_d = o Infr d = 1 Infr__d = O I
(gcf) _-
No financial .1033321 .1707017 .0851831 .14833
constraint
Minor fin. .1296224 -1767693 .1256266 .1848381
_Eonstraint /
Moderate fin. .2749876 .2969861 .3014354 .3294875
constraint i
I /
Major fin. .4920579 .3555429 .487755 _3373444
constraint

 

 

 

 

 

 

 

 

This methodology is now used to analyze the effect of the infrastructural dummy on
the ranking of the firms as regards financial constraints by comparing the different
probabilities in which infr_d = 1(firms have major/moderate infrastructure constraints)
with the situation in which infr_d = 0(firms have minor/no i“ffaStl'UCture constraints).
This methodology can be again implemented in two ways:

(i) In the first method, we start by making all firms in the sample have high

infrastructural constraints. We then calculate the individual probabilities under
this hypothetical situation. These probabilities are given in column 1 in table 6
and 7 under Probit and LOgit models respectively. Then we suppose all firms
in the sample have no infrastructural constraints, the individual probabilities

under this second hypothetical situation is given in cols 2 of table 6 and 7. The

82

 

difference between the cols l and 2 therefore the difference in the individual
probabilities with the infrastructural constraint “switched on” and “switched
off” respectively, with the value of no other variable altered.

(ii) In the second mflhOd, we compare the probabilities that result when the
dummy infr_d takes its two different values across all the firms in the sample,
with the values of other variables, in each case, held at their sample means.
This method thus compares the “straw firm” who apart from infrastructural
constraints embodies the average characteristics of all the firms in the sample
and who has infrastructural constraints in one scenario and no infrastructural
constraints in another scenario. The estimates of the four probabilities, of
having no/minor/moderate and major financial constraint can be compared
under the two hypothetical situations — first, when this firm has infrastructural
constraint and second, when it doesn’t (columns 3 and 4 of table 6 and 7).

As we saw in the earlier tables, two features of the tables 6 and 7 are significant. First, for
any two ways of computing the marginal effects there was not much difference between

the logit and the probit models. Second. for any one model, there Was considerable

 

difference in the probabilities calculated in the two different ways. The probability of the
firms having both no and major financial constraint was higher when the probabilities
were computed as a mean of the individual probabilities (method (i)) than when they
were computed from the average characteristics of the sample (method (ii)). This is
because in the latter method, the influence of extreme values is dampened when
individual values of the determining variables are set equal to sample averages. Barooh,

V.K (2001) suggests that the critical question is how the values of the other variables are

83

 

 

to be held constant, when the dummy Variable of interest takes two different values. The
second method assigns to each firm the values of the sample means. The common value
assigned could be the median, in which case there will be different outcomes. The first
method on the other hand calculates the mean from the individual probabilities and
therefore realizes a unique outcome under the two scenarios, in terms of the mean
probabilities.

Table 8: Finance as a constraint for Firms b difficulties faced as re ards
Infrastructure.

 

PERCIENEAGE OF FIRMS mama'suvms:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. No . Minor fin. Moderate Major fin .
f1nanc1al constraints fin . constraints
Constraints constraints
All firms 17.06 16.87 27. 83 38 24
All firms
having 7 . 5 12 .7 25
Infrastructur . S 53 O 3
a1
constraints
All firms
having no 19.2 17.6 29.4
infrastructur 33 O 8
a1
constraints

 

 

 

 

 

 

 

 

 

The table above shows the summary statistics regarding the Infrastructure-finance
linkage obtained from the data. As we see, row 1 indicates the ranking of all firms re.
Finance constraint ranking. Row two shows the probabilities among those firms who
have moderate/major ranking for infrastructure constraints (infr_d = 1) Row 3 shows the
probabilities for firms where the infrastructure dummy takes value infr_d = 0 (no/minor

infrastructural constraints). As we see, firms indicating that they major financial

84

 

constraints increases considerably When all firms are facing

infrastructural Con .
compared to when they are "0‘ (533% as compared to 33 87) 0 Sim“ as
. 0 . n

the Other
' ' ' l)
percentage of firms indicating that they and, the

h - .

ave no finanmal constraints decrea
s

considerably (7.5% as compared to 19.2%). es

All the tables above trace the effect that infrastructure has 0 h
n t e financi
al

constraints facing the firm by analyzing Coefficient on the d
urmny variable infr d T
— - he

probability of having a high/low ranking on the financial constraint was different for
. . . . each
firm havmg infr_d — 0 than for a firm havrng infr_d = 1. But this could mean that the
coefficients of every determining variable could be different in each of the two sub
considered. In other words, the equation should have been re-estimated all ' groups
owing for the
coefficients of all the other determining variables to be different in the groups when
infr_d = 0 and when infr_d = 1. In other ways, we have to allow for the fact that a" other
determining regressors (the region specific and firm specific variables) could be different
depending on whether the firm faces infrastructure constraints (infr_d = l) or whether the
firm faces no infrastructure constraints (infr_d = 0).
One way of doing this is to allow for “interaction variables” in estimating the
69 nation. We multiply the regressors by the infrastructure dummy. So, When infr_r = 1,
the coefficients attached to the interaction variables represent the additional contribution
t“ these coefficients resulting from the firm having infrastructural constraints, This is the

S i mgle equation “integrated” approach,

 

75116 second approach is to estimate two separate equations on the two subgroups one

\thh has infrastructural constraints (infr_d = l) and the other which has no

i hfrastructural conStraints (infr_d = 0). This second approach is summarized in Table 8

85

 

m I
which Shows teStfi‘rcw’e‘y the results 0f the lOgit estimations separately W/I
en

' - f d * 1 d 1717715 time
infrastructure constraints (1n 1‘..— ) an When they face no infrastmctupal
Constraints
(infr_d = 0)-

These regressions are shown below in tables 9 and 10, where the insignific
ant

variables were dropped. This is because their coefficients were individually and join t]
)’

not significant from zero. This 18 an Important methodological point: When equations a1-
e

used for prediction should they contain all the variables, even though some of th
e

coefficients may not be significantly different from zero or should they Contain Only
those variables with significantly non zero coefficients? One argument is that if one

believed a priori that a variable had a legitimate place in the equation specjfi

cation then
one should persist with this belief and include it. The other argument is that the purpose

and of estimation and prediction is to confront equation specification with data, to base
predictions on the coefficient estimates obtained from full specification may be

misleading since it would allow variables whose legitimacy in the specification had been

' 8
“rejected” by the data, to influence the predictions (Borooah, 2003). So, while Table

Summarized the estimations based on the whole set of variables, tables 9 and 10 Show the

Same based on a restricted specification.

It is interesting to see that in this restricted specification, some of the region

S‘hecific variables, namely the quality of central government (qgov‘d), the level of

i “nation in a country (infl_d) and the tax and regulatory policies (txreg\d) turn out to be

 

i “Significant when all firms are made to have infrastructural constraints, but they remain
§ ignificant when firms are made to have no infrastructural constraints, This means that

the effectiveness Of these variables on financial constraints facing the firm is nullified to a

86

 

 

 

great extent when firms face inf rash-“aural difficulties. Since Prov
‘ ISion of j
"fray;
”mu”?

is one of the basic factors determining the business environment f '
acrng the f
”7715, We

could 83)’ that the effectiveness Of these monetary and fiscal factors is l‘
r ea IZed on]
y When

the firms have fewer constraints With regard to infrastructure

Table 9: Ordered Lo it on the Rankin of firms re.
Sub-sam le of firms havin InfraStructural constraints infr d — 1

 
 

Ordered logit estimates Number of obs _
LR chi2(14) ; 1257
. Prob > ch12 = 126.28
Log likelihood = —1380.5618 Pseudo R2 7 8.0000
‘ 0437
f Coef. Std. Err. z p) ------------- ~
_____ g""95% Cour. Interval,
count 1 '08091416 02357151 3.43 0.001 . ------------ _
count::2 |.5023055 .1956456 2.57 0.010 .ii;;::; 1'271135
qgov_d |.1802941 .1163144 1.55 0.121 -.O47678 '8857637
1nf1_d |.1799364 .1365316 1.32 0.188 --0876505 '4082553
txreg_d |.2681226 .1739174 1.54 0.123 -.0727493 '4475334
gcorr_d |.5235881 .1210447 4.33 0.000 ~286344a '3232332
tadm_xegd|.3078607 .1291084 2.38 0.017 .0548129 :5609085
81201 |.6100626 .177939 3.43 0.001 .2613085 .9588167
31292 |.4637863 .1701533 2.73 0.006 .1302921 .7972305
lnvdebt |-.017755 .0075727 -2.34 0.019 -.0325974 —.0029129
fn_re |-.001653 .0014582 -1.13 0.257 -.OO45108 .0012051
inv_d |-.141776 .1176203 -1.21 0.228 -,372307 .0887561
debt_d |.3649487 .1289211 2.83 0.005 .1122679 .6176294
deb£_d |.2920326 .1265891 2.31 0.021 .0439225 .5401‘25
__ ___________ + _________________________________________________________
_Cutl I —.658983 .3449017 (Ancillary parameterSl
_cut2 1 .5433287 .3399581
_cut3 1 1.89135 .344221
gcf I Probability Observed
\ \_. ---------- +_ ____________________________________
130 obstacle I Pr( xb+u<_cut1) 0.0742
Ithz‘knor obstac | Pr(_cut1<xb+u<_cut2) 0,1255
mgderate obs I Pr(_cut2<xb+u<_cut3) 0.2644
”@3133: obstac \ Pr(_cut3<xb+u) 0,5359

 

87

 

      
 

Table 10: Ordered L0 '3 on the Rankin of firms re.
Sub-sam le of firms “3““ N o Infras‘l‘llctural constrai ' n3 ,-
Number of obs
2143

Pr(_cut3<xb+u)

Ordered logit estimates ‘
LR Chi2(14) ; 357 9
Prob > ch12 : . 1
Log likelihood = —2712.32‘43 Pseudo R2 7' g 0000
-------------------------------------------- ‘ 0619
Std Err. “““““““““““““““““
gcf | Coef. z P> z “
------------ -+”--”“”“'"’3;"‘““----—----1_1_____£?§% Conf_ Interval]

3 .173 58 3 91 ---------------
comm—1:1!1 l-5794“ - 0.000 .3392555 ‘
countryz |.6161416 .120823: 5.10 0.000 .3793327 1.019677

“and \.259 5773 .08323 3.12 0,002 .0964303 .8529505
in£1__d \.220 8901 .0912711 2.42 0.016 .0420”; .4227244
txreg_d \.697 3715 .100352 6.95 0.000 .5005353 .3997731
gcorr__d \.3493381 .0883 3.96 0.000 .1762732 -8940577
tadn regal .2763815 .0914586 3.02 0.003 .0971259 '522403
3712.1 \ .430622 .1461425 2.95 0.003 .1441331 455537
31292 .3915279 .1457874 2.69 0.007 .1057893 °Zl7055:
1nvdebt \-.o16495 .0057232 -2.88 0.004 -.0277122 - 6057;266
in re |-.001333 .0010599 -l.26 0.209 —.0034099 .000 778
“I”; |-.224072 .0843697 -2.66 0.008 -.3894337 -.05877:l4.47
debt_d .437958 .0948721 4.62 0.000 4.520121 .523”:
_____________ +--------——-------—---———--_--______________________-___-_
cutl 1 .3813447 .2302995 (AnCillary parameters)
cut2 1 1.394558 .2316537
cut3 | 2.770673 .2371747
-—— gcf I Probability Observed
_____________ +————----—--—---—-—-—-——-——----———-___
no obstacle | Pr( xb+u<_cutl) 0 1923
minor obstac | Pr(_cut1<xb+u<_cut2) 0 1759
inoderate obs | Pr(_cut2<xb+u<_cut3) o 2935
l 0 3383

Inajor obstac

3.5 Estimation over sub-samples - Characteristics versus coefficients

Table 8 tells us that 8% and 13% of the firms having infrastructural constraints had
1’) Q/minor financial constraints and 27% and 53% of them had moderate/major financial
Q Qnstraints. While 20% and 18% of the firms having no infrastructural constraints had
5 O/minor financial constraints and 29% and 34% of them had moderate/major financial
% onstraints. The fact that larger prepartion of firms having no infrastructural constraints

had no/minor financial constraints could be due to two reasons. First, all those

Qharactcn'stics Which increased the probability of a firm having high financial Constraints

88

 

d among those firms whiCh faced high infrastructural c
it”:
could be concentra

ODS/rams, or
. s which reduced the probability of a fi
teristic
those charac

rm having high

financial
concentrated among those firms Which faced low infras
' ts could be
constrain

tructurai
be a particular attribute which increased the probability of a firm
’ ts. Or may

constrain

‘ l constraint (being of a smaller size, present in a low-income country,
- ' h finanCIa
havmg hlg

. levels etc...) Were penalized more harshly if the firm faced
. high inflation
facmg

W w much of the financial constraint
ctural constraints. So, 6 have to find out ho
infrastfu - . . .
f 8 having infrastructural constraints and firms havmg no infrastmctufai
tween irm .
a? b6 . - - w much was it due to the
g s due to differences in characteristics and ho
' ts wa
constrain . . .
uestions.
in coefficients. We ask the followmg q
ifferences . o i . on S
d ld have been the predicted probabilities of the two different gr p
' What wou .
I) ' in them)
( ( h ving infrastructural constraints, and the other not hav g
a
of firms one . heir own
usmg t
' t' s ossessed by each group were evaluated .
if the charactens 1c p f fi S havmg
. ilities 0 1m
' '7 h t is we evaluate the predicted probab h
coeffic1ents. T a , 1 constraints eac
rm ' ' ructura
t ral constraints and fi 5 havmg no infrast
infrastruc u
at their own coefficients. t tural constraints
C l I S
. luate the probabilities of firms With low infra rue
" e eva
(ii) Then w .

' are
' St!‘ nts. These
ff ' ts of firms with high infrastructure con 3,
at the coe iCien
“synthetic” probabilities.
' ' ies.
..) We then compare the three probabilit . n
(in as 61‘ size. It gives a
f ub groups of firms, as per country and p
e is done or s -
.bhe sam

. . . f
' ° 6 1n the characteristics 0
d'fference in rankings being due to differenc
‘ . l
1 hdicatlon of the

89

 

 

- w’ .
We see that Whe“ firms 1th I0W infrastructure constraints w
CHCCHQIL'
a tcdaz their

own coefficients, we see that their Probability of having ho/mino /
r mOderate a
nd

- - - ' ts Were 19% 187
major fmancral 00%“th . o, 29% and 34% res '
Pectively. When th
6

characteristics Of these fi“ I ‘8 were eValuated at the coefficients of firms h ' h
avmg i gh
infrastructure constraints, the Pmbabiht)’ 0f having no/minor financial constraint f
S ell
to 10% and 15% respectively. The probability of having moderate fina
nCial

constraints remained more or less same at 28%, but the probability of having .
UDQJOr

financial constraints rose sharply to 47%. The story was repeated for the sub gro
- ups

considered with reSpect to the size of firms, and the country in which the 1‘”
runs

operate.

Table 11: Predicted Probabilities of Firms with Infrastructural Constraints and
Firms with no Infrastructural constraints havin different rankin

Legarding Financial Constraints.

PREDICT“ PROBABILITY OP HAVING:

 

 

 

 

 

 

 

 

No fin. Minor fin. Moderate W
Constraint constraint fin. constra n
constraint:
1111 Firm:
____________..._—
Firms with low
infrastructure 19.1 17.7 29,5 33.7

constraints at
their own
coefficients
Firms with low
infr. constraints
at the coefficients 9 .6 15 27.9 47.4
of firms with high
infr. Constraints
Firms with high
infrastructure
constraints at 7 .4 12 .7 26 , 5 53 .4
their own
coefficients

 

 

 

 

 

 

 
 
  
 

 

 

 

 

 

   

 

90

 

 

Small Sized firms

 

Firms with low
infrastructure
constraints at
their own
coefficients
Firms with low
infr. constraints
at the coefficients
of firms with high
infr. Constraints
Firms with high
infrast;ructure
constraints at
their‘own
coefficients

   
   
  
 
 
 
 

 

medium sized firms

y/r

financial
constraint

17.

NO M1301? fin .

coustraint

___________~

9

 

l7.

p——_____‘~_§__

No

financial
Constraint

 

13.

11.3

IMinor fin.
constraint

 

Firms with low

infrastructure

constraints at
their

own coefficients

18.

 

l7.

 

 

Firms with low

29.

 

in.

 

27,5

Mbderatg
fin,
constraint

M‘ior fin.
constraint

 

 

infr. constraints
at the coefficients
of firms with high
infr. Constraints

Firms with high

15. 28

a

 

infrastructure

constraints at
their own

coefficients

v

 

12.

26.5

 

34.5
47
) 53.8

 

Large 81'. zed firms

financial
Constraint

Firms with low

Nb

minor fin.

constraint fin.

Moderata

‘

major fin.
constraint

 

infrastructure
constraints at
their own
coefficients
Firms with low

27

.3

20. 28

 

infr. constraints
at the coefficients
of firms with high
”_£g§£;JBEE§Eraints

15.

 

20. 29.

 

 

 

91

constraint

 

 

24.

34.

 

 

  

‘zomltnthnt

 

Firms with high
infrastruCi—ure
Constraints at.
their own
coefficients
71m in low income
countries

 

infrastructure
constraints at
their own
coefficients
Firms vaith.low
infr. constraints
at the coefficients
of firms ‘with high
infr. Constraints
Firms with high
infrastructure
constraints at
their own
coefficients

/

 
 
   
   
 
  
 

 

f
Firms wi th low

N

Constraint

12.2

18.1

Nb
financ 13,1

16.2

 

\
Minor fin .
constraint

 

30.2

MOderata
fin,
constraint

 

 

 

 

Firms in medium
income countries

financial
Constraint

N° Minor fin .

constraint

29.9 -
25.4 .
23.7 -
M(Marat;

“€301: fin.
fin.

constraint

 

 

Firms with low
infrastructure
constraints at
their own
coefficients

16.

°°n8traint

 

Firms with low
infr. constraints
at the coefficients
of firms with high
infr.

 

 

30.3 36.8
27.7 L

 

 

 

Firms with high
infrastructure
constraints at
their own
coefficients

26-7 ) 53.5

 

Firms in high
income countries

No
f inancial
Constraint

Minor fin .
constraint

Moderate

Major fin.

fin ' constraint

 

FirmS‘with low
infrastructure
constraints at
their own
coefficients

  

 

35.

1

constraint

 

24.5

16.

 

 

 

Firms with low
infr. conStraints
at the coefficients
of firms with high
infr. Constraints

  
  

 

 

18.

 

 

30.1 28.

 

 

 

92

M4110.1:- fin .

 

 

     
   
 
     

Firms with high
infrastructure
constraints at
their Own
coefficient“S

With no exceptions we find that With reSpect to all sub

'8f0UpS, the probability of having
no financial constraints always fell and the probability 0f having major financial

constraimS rose when the firms han g low infrastructure constraints were evaluated
at

the coefficients of the firms with having high infrastructure

constraints, though the
magnitude of these changes always varied as per the sub-group being considered. In the

case of firms as per the countries in which they were situated, this pattern was sharper for
firms in the high-income countries. This was in contrast to the films in low-income
.countries, which were more varied in terms of spread of countries. With respect to firm
size, the sharpest fall (in the probability of having no financial constraints) and the
sharpest rise (in the probability of having major financial constraints) were in the case Of
the small-sized and medium sized firms. This definitely reinforces the analYSiS that it is

the smaller and medium sized firms who benefit most from good infrastructure.

3.6 Conclusion

.bliis empirical study re-enforces the link between infrastructure and private sector
D‘Oductivity. One of the crucial aspects of private sector productivity is the ease with
which they can borrow and fulfill their contracts in the capital market. The quality of the
infrastructure facing the firm crucially determines the difficulties they face as regards

fi mance. By shOWing that firms facing high infrastructural constraints are the ones most

93

 

likely to face high financial constraints as Well .
, this study poin
ts out to the, float/,3: the

. . ‘t 'nf .
PTOVISIOH 0f 300d quall y I rastrucmre COUSIderably eases the Problem of getting finance

for firms. By analyzmg firm level.

ua ' ' . . -
q litative data this study points out to a major l1nk
. - r .
through Which 1ntrastmcm e: ca" affeCt pnvate sector product 't
1V1 y.

While this 8‘10““ “0‘ be seen as one of the solutions by whiCh the financm'
constraints of the firms can be ameliorated, it does go a long way in pointing Out the

“spill-over” benefits 0f building good infrastructure. This is especially crucial in the case

of small-sized firms who are resource constrained and who will derive the maximum

benefits from good quality public infrastructure. In the Case of high—income countries.

this infrastructure-finance link is more potent as compared to the low-income countri65~

 

3.A Appendix : Definitions of variables

Dependant variable : gcf — General constraint _ financing (no obstacle; minor obstacle;

moderate obstacle; major obstacle)

Independent variables:

- Region specific variables influeming financial constraints faced by a firm:

1.

2.

countryl - low income country dummy

country2 - middle income country dummy

qgov_d - Dummy for quality of central govt. in a COllntry (1 = quality \ slightly
bad to very bad; 0 = quality very good to slightly good)

gcpi_d - Dummy for political instability in a Country (1 = political instability is a
major/moderate constraint; 0 = political instability is no/minor constraint)

infl_d - Dummy for inflation in a counny ( 1 = inflation is a ma'lorlrnodetate
constraint; 0 = inflation is no/minor constraint)

txreg_d - Dummy for taxes and regulations in a country (1 = taxes and regUlationS
are major/moderate constraint; 0 = taxes and regulations are a no/minoi-
constraint)
gcorr_d — Dummy for corruption levels in a Country (1 .._.

CoITuptjoH is a

major/moderate constraint; 0 = corruption is no/minor constraint)

tadm_regd — Dummy for tax-administration regulations (1 = tax~adm
' r CSUIations

are a major/moderate constraint; 0 == tax-adm. regulations are a
nO/mino
I'

constraint)

95

 

 

 

9. infr_d - Dummy for infrastructural constraints ( l = infi‘astmcmm is a

major/ moderate COUStfaint; 0 = infrastructure is no/minor constraint)

‘ Firm specific variables influencing financial constraints faced by a firm:
10. sizel - Dummy for small firms

1 1. size2 - Dummy for medium films

12. lnvsal — log of the value of sales(vsal) ($)

13. lnvfas — log of the value of fixed assets (vfas) ($)

14. lnvdebt — log of the Value of debt (vdebt)

15. fn_re - source of investment finance — retained earnings ( 1 =

yes; 0 = no)

16. sal_d - dummy for increase in sales over the past 3 years ( 1 = yes; 0 = no)

17. salf_d - dummy for expected increase in sales the next 3 years ( 1 = yes; 0 = n0)
. . . . . O :5 “0)
18. 1nv_d - dummy for increase in investments over the past 3 years ( 1 =5 363’
' ° - st 1 5 YES;
19. invf_d - dummy for expected increase in investments over the next 3 Y‘
O = no)
20. debt_d - dummy for increase in debt over the Past 3 years ( 1 == yes ; O = no)
21. debf d _ mm for expected increase in debt the o t
_ du Y ver the nex 3 years ( I =yes;
O = no)
22. lab_d - dummy for increase in employment over the past 3 years ( I s
23. labf_d - dummy for expected increase in employment over the next
3 Years ( 1 s.

yes; 0 = no)

24, afs_d — dummy for audited financial statements ( 1 = yes-

O=no)

96

 

Bibliography

[1] Aghion Phillipe and Bolton, Patrick. “A Theory of Trickle-Down Gro
’ - . - Wth and
Development” Review of Economzc Studzes, 84 Issue 20997).

[2] Akerlof, George, «The market for lemons : qualitative uncertainity and the ma
ket mechanism” Quarterly Journal of Economics, Vol. 84 (1970). r-

[3] Ballard, Charles L" and Fullerton, Don. “Distortionary Taxes and Provision
of Public Goods” Journal ofEC‘momw PeTSpectz'ves Vol.6 No.9 Summer 1992.

[4] Bardhan. Pranab “Corruption and Development: A Review of Issues” Joumal of
Economic Literature Vol.35 No.3 Sept. (1997),

[5] Batra, Geeta; Kaufman, Danieland Stone, Andrew “Voices of Fir ms, :200)? -
9 Findings of the World Business Envnonrnent Survey” World Bank (forthcomlng

u - Destrucl
[6] Baumol, William J. Entrepreneurship: PFOduCtiVe, Unproductive and

tive” Journal of Political Economy V0198 Issue 5 Oct.(1990)-

,. - ificance
[7] Berger, A. N and Udell, G. F. “Some Evidence on the Empiricifi Eng“

of Credit Rationing” Journal of Political Economy Vol. 100 No. 5 (1992).

[8] Bernanke’ Be" and Blinder Alan 3' “Credit Money and Aggregate Demand”
American Economic Review, (1988)-

ler Mark. “A enc C W
[9] Bernanke, Ben and Gert a. . g y osts, Net Orth .
Fluctuations” American Economic RC'UZC'LU, Vol. 79, (1989). and Busmess

[10] Bernanke, Ben and Gertler, Mark. “Financial Fragility and Eco; .
malice” Quarterly Journal Of EconomiCS, Vol. 105, Issue 1 (1990). 10mm Perfore

[11‘ Bernanke, Ben and G ertl er, Mark. .tInside the Black Box : The Cr (1‘
of Monetary Polciy Transmission” Journal of Economic Perspectives 8 It Channel

[12] Bernanke, Ben; Gertlel‘, Alan S. and Gilchrist, Simon. “The Fin

celerator and the Flight to Quality” Review of Economics and Statisttincial AC~
No. 1 (1995). ’ V01 78,

[13] Bernhardt, Dan. “Credit Rationing?” 247’1'67'160" Economic Review Vol. 90 No 1
(2000). ’ '

97

 

 

[14] Besley, Timothy. “How Do Market Failures Justify Interventions in Rural Credit
Markets?” World Bank Research Observer, Vol. 9 (1994).

[151 Besanko’ D and Thakor, A. “Collateral and Rationing : Sorting Equilibria in
a Monopolistic and Competetive Credit Markets” International Economic Review
Vol. 28 Issue 3 (1987)-

[16] Bester, H- “Screening Versus Rationing in Credit Markets with Imperfect Infor-
mation” American Economic Review, Vol. 75, ISSUe 4 (1985).

[17] Binswanger, et el. “How Infrastructure and Financial Institutions afiect A .
. (1 n J mat D ~ grlCUI-
tural Output in In 1a 011 of evelopment Economics, Vol. 41 (1993) _

[18] Binswanger, H.P. and Rosenszweig, Mark. “Behavioral and Material d
minants of Production Relations in Agriculture” Journal of Development Stuff?“
Vol. 21 (1986). m.

[19] Blinder, Alan S. and Stiglitz, JO_Seph '13- “Money, Credit Constraints and
Economiv Activity” American Economzc 361128111, May (1983)

[20] Boadway, R'W" and Keen, M' “Pflblic g°°dS, Self Selection and Optimal 19’
come Taxation” International Economic Review, Vol. 34 (1993)

es for Be

(21] Boadway, R.W., and Marchand, M- “The Use of Public Expenditm'

distributive Purposes” Oxford Economic Papers V0]. 47 Issue 1. (1995)
, 11 GOING-(flan

[22] Bo ad way, Robin «Public economics and the theor of ublic pofic‘i
Journal of Economics Nov. (1997). y p
M06815” sag e

[23] Borooah, Vani K “Logit and Probit: Ordered and Multinomi 07 138

University Paper Series on Quantitative Applications in Social Sciences:
Thousand Oaks, CA. Sage.(2001)

[24] Boyd, John H. and Prescott, Edward C. “Financial Intermed' . .
Journal of Economic Theory Vol. 38 (1986). 1313' Coahtxons»

[25] Buiter, Willem H “From Perdition to accumulation? The Se
decade in Russia” Economics of Transition Vol.8 No.3, (2000) COnd transition

[26] Clemenz, Gerhard. “A Note on the Non-Existence of Rationin E ,
the Besanko-Thakor Model” International Economic Review, V01, 35 I qmlibrjum in
i SSHe 3 (1993)

[27] Dandekar, V.M. “Limits of Credit. not Credit Limits” Economic a a’
” Politica
1

Weekly, Sept 25 (1993).

[28] De Meza, David and Webb, David. “Too Much InveStmem .
Asymmetric Information and Incomplete Markets” Quarterly JOurnal
Vol. 102, Issue 2 (1987).

A P roblem of

of ECOnomics,

98

 

 

[29] De Meza, David and Webb, David. “Credit Market Efficiency and Tax pone},
in the presence of Screening Costs” Journal of Public Economics, Vol. 36 (1988)-

[30] De Meza, David and Webb, David. “The Role of Interest Rate Taxes in Credit
Markets with Divisible Projects and Asymmetric Information” Journal Of Public
Economics, Vol. 39 (1989)-

[31l De Meza, David and Webb, David, “Risk, Adverse Selection and Capital M ar-
ket Failure” The Ewnomw Journal, Vol. 100 Issue 399 (1990). ,

[32] De Meza, David and Southey, Clive. “The Borrowers’ Curse : Optim‘
Finance and Entrepreneurship” The Economic Journal, Vol. 106, Issue 435 (19153,

[33] De Meza, David and Webb, David. .cDoes Credit Rationing Imply Insuflicie t
n

Lending?” Journal of Public Economics, Vol. 78 (2000).
[34] De Meza, David. “Overlending?” The Economic Journal, Vol. 112, No. 477 (2002) -

[35] Diamond, D W- “Financial Intermediat ion and Delegated Monitoring” Review Of
Economic Studies Vol.51, Issue 3 (1984).

i36l ESfahani, H S. and Ramirez, M 1" “Institutions, infrastructure and economic
growth” Journal of Development ECOnOmics, Vol.70 (2003)

. ° . . . . 1/
[37] Feehan, James P. “Public Investment ; Optimal Provision of Hicksla“ public If

puts” Canadian Journal of Economics V0131 No.3 (1998). ed' t.

. . - 'n C‘ 1

[38] Gale, William G. “Collateral, Rationing and Government Intervefi‘wnn; NBER.
Markets” in Asymmetric Information, Corporate Finance and 112.065th ’

(1990).

[39] Galor, O. and Zeira, J. “Income Distribution and Macroeconomics” Review of
Economic Studies, Vol. 60 (1993).

[40] Gertler, Mark. “Financial Structure and Aggregate EconomiC A

' C . . ,,
of Money, Credit and Banking, Vol. 20 Issue 3 (1988). “WW Journal

[41] Gertler, Mark and Gilchrist, Simon' “Monetary Policy, Blisine.
Behaviour of Small Manufacturing Firms” Quarterly Journal of Eco, ‘53 CyCIes and
(1994). ”mics, Vol. 19

[42] Globerman, Steven and Shapiro, Daniel “Global Investment a, 1
vestmcnt flows: The role of Govcrnancc Infrastructure” World Develo 1( Direct In-
No.11 (2002). Men: Vol 30

[43] Gramlich, Edward M .‘Infrastructure Investment: A Review Essay” J
Economic Literature Vol. 32. Sept. (1994). Ournal of

99

 

 

[44] Greenwald, Bruce and Stiglitz, Joseph E. “Externalities in Economics with
Imperfect Information and Incomplete Markets” Quarterly Journal of Economics
Vol. 101 (1986).

[45] Hall, Robert E. and 301188, Charles 1. “Why do some countries produce so
much more output per worker than Others?” Quarterly Journal of Economics Feb

(1999).

[46] Hellwig, Martin. “some Recent Developments in the Theory of Competing“ in
Markets with Adverse Selection” European Economic Review, Feb/ Mar (1987)

[47] Hoff, Karla; Braverman, Avishay and Stiglitz, Joseph E. “The Econ .
of Rural Organization, Theory Practice and Policy” World Bank (1993). omcs

[48] Hoff, Karla “The second theorem of the second best” Journal of Public Economics
Vol. 45, (1994).

[49] Hoff, Karla and Lyon, Andrew B- “N Oil-Leaky buckets: Optimal I‘edistributive
taxation and agency costs” Journal of Public Economics Vol, 58, (1995),

[50] Hoffman, Mathias “Cross-country evidence on the link between the level of irl’
frastructure and capital flows” Applied Economics No.35 (2003).

. . 7'1 6
[51] Holtz—Eaken, Douglas “Public Sector Capital and the Productivity Puzzle Th’

Review of Economics and Statistics Vol. 76 No.1 (1994).
u - 011,611“ 9
[52] Jaffee, D and Stiglitz, J .153. Credit Ratiomngn in The Han dbool‘ OfdM

Economics. Vol II. B.M Friedman and F, H Hahn (eds). North Holmn i

mdam. (1990).

[53] Leland, H E. and Pyle, D H- “Information AysmmetrieS, Financial Structure
and Financial Intermediatlon” Journal of Finance, Vol.32, Issue 2 (1977),

[54] Levine, Ross. “Financial Development and Economic Growth :

. . Views
Agenda” Journal of ECO‘ROm’lC Literature, V01. 35, Issue 2 (1997) , and
[55] Mankiw, Gregory, “The Allocation of Credit and Financial C01 la ”
Journal of Economics Vol.101 ISSUE! 3 (1986). p Se Quarterly

[56] Mauro, Paulo. “Corruption and Growth” The Quarterly Journal
Vol.110 Issue 3 August (1995)- of Emmmz'cs

[57] Munnell, Alicia H.(ed-) “Is There a Shortfall in Public Capital I
Conference Series No. 34 Federal Reserve Bank of Boston. June 1990 “Vestment?a

158] Munnell, Alicia. “Policy Watch ‘ Infrastructure InVeStlnent and

Growth” Journal of Economic Perspectives Vol.6, No.4. Fall 1992. EConomic

100

 

 

[59] Parker, Simon C. “Do Banks Ration Credit to New Enterprises? AndSbould Gov-
ernments Intervene? Precident’s Lecture Delivered at the Annual General Meeting

of the Scottish Economic Society” Scottish Journal of Political Economy V01- 49
No. 2 (2002). y

[60] Rothschild, Michael and Stiglitz, Joseph E. “Equilibrium in Competet'we

Insurance Markets An Essay 011 Economics of Imperfect Information” Quart erly
Journal of Economics, Vol. 90 (1976).

[61] Smith, Bruce D and Stutzer, Michael J_ «Credit Rationing and Coven“,n
Loan Programs = A Welfare Analysis” AREUEA Journal, Vol 17, No. 2 (1989)hem

162] Spence, Michael “Signalling in RetrOSpect and the Informational Struct
Markets” American Economic Review, Vol. 92, No. 3, June (2002). ure of

[63] Stiglitz, Joseph E. “Interest Rates, Risk and Imperfect, Markets : P
Policies” Oxford Economic Review, Vol.15, No.2 (1999). Uzzles and

[64] Stiglitz, J. E. “Information and the Change in the Paradigm in Economics” Ame?”-
ican Economic Review, Vol. 92, No. 3i J me (2002).

[65] Stiglitz, Joseph E and WRSS’ Andrew “Credit Rationing in Markets with
Imperfect Information” American Economic Review Vol, 71 Issue 3 (1981).

' t
. . ' “ . - Cred-1
[66] Stiglitz, Joseph E. and W8188, Andrew. Asymmetric Information InVOl- 44..

Market and its implications for Macroeconomics» Oz for d Economic Papers,

Issue 4 (1992).
- tionin ”
[67] Waller, C J. and Lewarne Stephen. “An EXPOSitory Model of C Ted“ R3 g
Journal of Macroeconomics, Vol. 16 No. 3 (1994).

[68] Wei, Shang—J in “L0call Corruption and Global Capital Flows” Brae/ting? Papers
on economic Activity V01 2 (2000)-

[69[ Wilson, Charles. “A Model of Insurance Markets with Incom ,
Journal of Economic Theory, Vol.16 (1977). p1 ete Information n

[70] The World Bank Group “The World Business Environment
http: / /info.world bank.0rg / governance / wbes /. SUFVey, 2000. ,,

[71] World Bank, (ed.) “ Infrastructure for Development: 1994 V'VOrI
Report.” Washington DC: WOTId Bank, 1994. d DCVGIOPment

101