regulation and regulatory reforms in developing countries antonio estache european school on new...

40
Regulation and Regulatory Reforms in Developing Countries Antonio Estache European School on New Institutional Economics ESNIE 2009 Cargèse May 2009

Post on 21-Dec-2015

214 views

Category:

Documents


0 download

TRANSCRIPT

Regulation and Regulatory Reforms in Developing Countries

Antonio Estache

European School on New Institutional Economics ESNIE 2009

Cargèse May 2009

Overview• Focus on regulation in key

infrastructure industries– Some background data on the main

reforms of the last 15 years

– A zoom on regulatory reforms

– A further zoom on the institutional dimensions of regulatory reform in developing countries

The reforms of the 1990s• Three main “standard” reforms:

– Relying more on competition• In the market when possible• For the market otherwise

– End to old fashion self-regulation when regulation was still needed

• create “independent” regulatory agencies• deal more explicitly with the incentives for efficiency

in the design of regulation (i.e. replacing cost + by price caps)

– Opening up to private sector to get access to private financing to fasten service coverage increases

Mixed to poor success of the efforts to attract the private sector…

% of countries with Private Participation in Infrastructure (2004)

Elec.Gn. Elec.Dis. W&S Rail Telecms

Tot. Dvlping

44% 36% 35% 37% 48%

Tot. Dvloped

70% 43% 80% 65% 83%

Total 48% 36% 42% 43% 55%

Investment commitments to infrastructure projects with private participation in developing countries in real and nominal terms, 1990–2007

0

30

60

90

120

150

180

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

US$ billions 2007 US$ billions

Total: US$1,475 billion committed through +/- 4,100 projects… adds up to less than 20% of the investment in the sector….

144

111

158

Source: World Bank and PPIAF, PPI Project Database.

Telecoms and energy dominate investments levels

0

30

60

90

120

150

180

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Energy Telecoms Transport Water and sewerage Total

2007 US$ billions

Source: World Bank and PPIAF, PPI Project Database.

Energy and transport dominate the number of projects

0

50

100

150

200

250

300

350

400

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Energy Telecoms Transport Water and sewerage Total

Source: World Bank and PPIAF, PPI Project Database.

Projects

East Asia and Latin America are the favorite destinations

0

10

20

30

40

50

60

70

80

90

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

East Asia and Pacific Europe and Central AsiaLatin America and the Caribbean Middle East and North AfricaSouth Asia Sub-Saharan Africa

2007 US$ billions

Source: World Bank and PPIAF, PPI Project Database.

0

20

40

60

80

100

120

140

160

180

1990

1992

1994

1996

1998

2000

2002

2004

2006

2008

0

1

2

3

4

5

6

7

8

9

PPI GDP growth (annual %)

PPI in infrastructure et PIB growth

2007 US$ billionsPercentage

One slide on the crisis and infrastructure investments…not good news…

-200

-100

0

100

200

300

400

500

600

700

800

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008e

2009f

Equity investment, net Commercial banks, net

Private flows to LDCs forecast

US$ billions

Source: World Bank and PPIAF, Impact of the financial crisis on PPI database, PPI Projects database, and Institute of International Finance.

The regulation independence war also only enjoyed only a mixed success(% of countries with Independent Regulatory Agency)

Elect. W&S. Rails Telecms

Total Developing 50% 26% 7% 61%

Total Developed 92% 20% 21% 55%

Total 57% 25% 9% 60%

In a nutshell….How did it work out?

• Fiscal cost: ok in short term but not ok over the longer run

• Efficiency: reasonably ok although increased shift from price caps to hybrids dominated by strong costs pass thru rules

• Equity: key problem and essentially a regulatory design issue

• Accountability: not good either…and again a regulatory design issue

The winners and losers in terms of Actors• The actors in the payoff matrix

– The users (access: (+ but not as much as expected and distributional issues), affordability(-), quality (+))

– The taxpayers (cash!: + in SR, -/+ in LR)

– The workers (jobs + cash: - in SR, + in LR)

– The operators (cash in the SR and IRR> COC in the LR for a few! (+ in SR, ? for LR)

– The local owners (cash! + in SR and LR)

– The foreign owners (cash! + in SR, +/- in LR)

– The bankers (cash! + in SR and LR)

– The politicians (cash! + in SR and LR)

– The donors (???)

Emerging Issues• For users:

– Residential users: Distributional issues– Non-residential users: could do much better

• For operators:– Demand uncertainty– Cost levels revisions to address overruns– Projects design revisions– Exchange rate risks and other economic shocks– De facto expropriation risks

• For government:– Uncertainty about demand and costs!– Uncertain net fiscal effects– Fiscal space: the illusion of private sector invest– Weak regulatory capacity, commitment and strong

capture• But counterfactual may be worse!

How can regulation theory help in the diagnostic

and …how can it help fix

things?

First: recognize a few basic principles emerging from theory…

1. Information asymmetry matters and can matter a lot!

• When a regulated operator has privileged information: it usually gets a rent from it

• Information asymmetry seems to be a much bigger issue in LDCs

2. Rents exist…but are not necessarily bad!• Regulation can be designed to get operator to use the rent (within

limits…) in a Pareto improving way

3. Limited commitment ability of a regulator is an essential driver of the effectiveness of regulation in terms of efficiency, equity and fiscal costs

• It may be a good idea to limit a regulator’s powers to avoid undue use of information through capture associated with a limited ability to commit

15

How can these issues be modeled?• Ideally, we need a general model to see how a

monopolist will behave to maximize rent from weak institutional capacity

• But we need to make sure that the main institutional capacity issues generally recognized by experts on LDCs can be addressed explicitly within the model

• This also means we need to explicitly separate the regulator from the government to check for regulator specific problems

• In addition, we want try to reconcile the policy recommendations emerging from research focusing on narrow issues using issues specific models

16

So what are the institutional weaknesses we need to track down?• A survey of policy and theoretical

literature identifies:– Limited capacity/skills to regulate– Limited accountability– Limited ability to commit– Limited enforcement capacity– Limited fiscal efficiency

Each of these dimensions needs to hit on a specific variable in the general model

17

Some basic stylized facts on these institutional weaknesses? (1)

• Limited capacity to regulate or to enforce– Regulators are severely under-resourced …– … which can lead to increased firm rents.

• Limited commitment– Many contracts have been renegotiated…– … tends to increase the cost of capital…– … but could this effect have been decreased by

independent regulation…– … and better checks and balances???

18

Some basic stylized facts on these institutional weaknesses? (2)

• Limited accountability– Regulators (and governments) are often not

accountable… – … which decreases efficiency and inequality

• Limited fiscal space– High cost of public funds…– … partly explains why SOEs have not expanded

network enough…– … but increasing access is not profitable for

privatized firms…– … partly because of conflicts between affordability

and access…– … independent regulation appears to help

19

A Basic Model of Monopoly Regulation:(1)The Monopolist

• Monopolist produces a quantity q of a good with a fixed costs of F and a marginal cost of C(q)

• Monopolist cost-function is also driven by:– e = firm effort (i.e. moral hazard variable)

• Exerting effort e causes firm disutility of ψ(e) (ψ’>0 , ψ’’>0 , ψ’’’ ≥ 0 )

=> this is the controlable part of costs– β = underlying cost outside of firm’s control (i.e.

adverse selection variable such as technology or factor prices=> this is the uncontrolable part of costs

• β= β (low cost) with probability v, or high cost with probability 1-v.

C(q) = (β-e)q + F• Monopoly utility is U = qp - (β-e)q – F – Ψ (e) + t

With p=price and t=transfer

• Participation constraint: U>0 (once β is revealed)

20

• Consumer welfare:• S(q) = gross surplus =

• P (q) = inverse demand function

• λ >0 is the opportunity cost of public funds• Consumers maximize welfare p=P(q)=S’(q)

(3) Government• Benevolent government welfare function:

• Government always observes F & c = (β - e) • !!!but government does not observe not β or e (i.e.

composition of cost)• In order to learn β and e, the government employs a

regulator…

A Basic Model of Monopoly Regulation:(2) Consumers

0( )

qP q dq = ( ) 1 )(V S q qp t

= = ( ) ( () )W U V S q e eq F t

21

A Basic Model of Monopoly Regulation: (4) The Regulator

• Intuitively, the idea is that the main focus of the regulator is the cost function and the variables it can control β and e

• The firm’s cost β is revealed to the regulator with probability ξ

• If the regulator learns β, it chooses whether or not to reveal it to the government– Signal is `hard information’ – i.e. regulator cannot report a cost-

level that it has not observed• Government can incentivise regulator to reveal signal by

paying s if β is revealed– Social cost is λs due to cost of public funds

• Firm can incentivise regulator to hide signal by paying bribe– Such side-transfers are illegal and hence costly=> regulator

receives only a fraction 0 < k < 1 of bribe • We assume the gvt decides on the set of contracts offered

to the firm BEFORE the regulator makes its report on costs => gvt can influence regulator’s choice on info revelation

22

Complete Institutions Benchmark:Symmetric information

• In this version, the country does not suffer from any of the 4 institutional weaknesses

• If regulator reveals β, there is symmetric information Government maximizes welfare W, s.t. binding participation

constraints (PC)dW/dq=0 leads to usual markup price over mrgnal costs

( = elasticity of demand)

=>since for a given β, the only variable the regulator can focus on is effort (e) => dW/de=0 to get the optimal effort the gvt aims which leads to

- ψ’(e)=q (i.e. efficient effort) - U=0 (i.e. no rent)=> Price is set above MC to cover for cost of transfers which is

itself set to avoid any rent and H or L firms efforts are both optimal

( ) 1

1

p e

p

23

Complete Institutions Benchmark: Now…Asymmetric information (1)

• If regulator does not reveal β: Asymmetric information For a firm to be interested in any offer by the gvt, the offer

has to satisfy the incentive compatibility constraints (ICC)

• Here and

• i.e. firm is incentivized to reveal β truthfully Std result: the binding PC is for high cost firm and ICC is for

low cost firm Low cost firm is given positive rent: ; • an information rent which does not apply to high cost firm

( )

( )

U U e

U U e

( ) ( ) ( )e e e

( )U e

24

Complete Institutions Benchmark:Asymmetric information (2)

Now…if regulator reveals information, low cost firm gets no rent Low cost firm has incentive to bribe the regulator to keep β hidden Willing to bribe regulator up to Government is willing to pay conditional transfer to regulator to

prevent the regulator accepting bribe, i.e. • Regulator only get k with 0<k<1 due to transaction costs• Paying the regulator allows the gvt to avoid a cost to society of λk

; rent given to firm is costly to society since there is opportunity cost of public funds

Now we can calculate the optimal gvt choice of q and e Do so by computing the expected welfare E(W) given ξ and v dE(W)/dq=0 and get usual markup formula for price And…dE(W)/de=0 and get a complex formula

(2)

( )e

( )s k e

( ) 1 ( )1 1 1

ve q k e

v

( )e

( )e

25

Complete Institutions Benchmark:Asymmetric information (3)

What does this equation mean????,

Marginal disutility to effort of the firm (ψ’) can be impacted by a set of variables of relevance to the government

This includes the fact that the rent that the low cost firm receives is costly to society (comes from distortive taxation)

The gvt wants to minimize it To reduce the rent, the gvt can make the high cost

firm production (q) level less appealing to the low cost firm (work on e and hence Ø(e))

( ) 1 ( )1 1 1

ve q k e

v

26

Complete Institutions Benchmark:Asymmetric information (4)

We end up with an actual level of effort is lower than the efficient one

The 2nd term is increasing in v since the more likely the firm is to be low cost, the less likely the distortion in effort will occur and hence the gvt can allow the distortion to be greater

Note: gvt can act directly on cost looking at effort BUT it can also introduce an incentive scheme to get the firm to do the right thing on its own

Rather than setting p, cost or transfers, the gvt can set the price and come up with a reimbursement rule which makes the firm decide on the optimal effort level to max the rent (from low powered to high powered)

Note: in this particular model, the level of incentive is equivalent to the level of effort

27

Now we have a model…so what?

Let’s use the model to review the impact of each institutional weakness and the optimal policy response to each weakness

Let’s see how consistent these various optimal policies

Let’s see how these optimal policies match the standard policies recommended and often adopted by regulators in developing countries

What if limited regulatory capacity?• Limited regulatory capacity implies

– (a) lower ξ , the proba that the regulator observes the firms’ type or

– (b) no observation of C = (β - e)

• So what?– (a) From equation (2), lower ξ implies higher powered

incentives needed since collusion btw firm and regulator occurs less often and hence anti-collusion payments less of a concern

– (b) Non-observation of c implies high-powered incentives by definition – price caps are the only option

• => less capacity makes a stronger case for high powered incentive regulatory regimes

29

What if limited accountability? (1)• Less accountability of the regulator can imply greater value

of k (the cost to the government to cut the ease of making bribes)

Less likely to be optimal to prevent capture because so expensive to do so

Lower social welfare and greater frequency of capture

• Assume that with probability ζ regulator is `dishonest’ and will take bribes and with probability 1- ζ is `honest’ and will not …but the gvt does not know the regulator type

Strange result…with less accountability…may no longer necessarily be optimal to prevent regulator’s capture through payments to this regulator!

30

What if limited accountability (2)• Now what if the problem of limited accountability is not about

the regulator but about government non-benevolence?

• Less accountability of government can thus be modelled as

greater value of γ => can be modelled as misaligned objective

function and favouring of firm over consumers: W=V+γU• Government favoring of firm implies cares less about rent,

hence lower distortion of effort: • Now E(W) also includes γ and dE(W)/de=0 leads to drivers of

marginal disutility to effort of the firm (ψ’) as follows:

greater γ results in higher powered incentives to cut costs =>but associated with higher risks of regulatory capture…

( ) 1 ( )1 1 1

1 ve q k e

v

31

What if limited accountability (3)

• So do we have any solutions???• 1. recognize that limited accountability is mainly

due to lack of information flows between actors!• 2. this means that we need to reduce the

importance of information that any agents holds• 3. this can be achieved for instance by

– Lowering the power of incentives (cost plus looks good!)

– But also by the creation of new information sources (get multiple regulatory agencies to generate competition for the generation of information)

• …but not easy if you have limited capacity…

32

What if limited commitment (1)• 3 forms: (i) too much renegotiation, (ii) non respect of promises to firm

and (iii) limited enforcement willingness• => inefficiency ex-post and all firms will pretend to be high costs and

hence gvt needs of give up more rent to get the right ones on board• => high risk when need to make long term investments

• If firm invests I to influence β, it increases the probability that it will be a low cost firm i.e. ν= ν(I) (ν’>0, v’’<0)

• If government can commit to rents at time of investment, will set firm’s payoffs to account for all surpluses as follows:

• However, if no commitment, firm only considers private payoffs, hence

• =>Limited commitment therefore implies under-investment• NOTE: it will also lead to “ratchet effect” (if the firm reveals

its type to be low cost, it knows the gvt will be more demanding => added incentive NOT to reveal information => gvt could increase welfare by promising not to use info!

'( ) 1/ U V Vv I U

'( ) 1 U Uv I

33

What if limited commitment? (2)

• So…solutions?• Note again: gvt led renegotiations are common (unhappy with high

firm profits) => odds are driven by size of rent• => government can only commit to give a maximum expected rent

(let’s call it c) • If not satisfied, gvt needs to reduce e • => To satisfy this ICC, the gvt may have to favor lower incentives!• This is because the threat of renegotiation constrains its ability to

offer the firm the possibility of making large profits!

• =>in practice, this means that limited commitment may require also a reduction in power of incentives since need to give more rent than it otherwise would have to get the firms to participate in the business

• NOTE: empirical evidence suggests that price cap are more associated with renegotiation than cost-plus

( )e c

34

What if limited commitment? (3)

• More solutions?– Nationalization…since gvt end up happy with

the rent they now control…=> tolerate higher profits!

– Increase debt financing since gvt has to tolerate more interest payment than it tends to tolerate dividends

– Increase independence of regulator

• The fact is that each form of lack of commitment tends to lead to its own solution!

35

What if limited fiscal efficiency? (1)• If no money for direct subsidies…a natural solution are

cross subsidies (across people or across regions)• Consider two regions – rich (1) and poor (2)• In region 2, only a share θ of the population are connected.

• Let Fi , ci , qi , pi be the fixed cost, marginal cost, quantity per capita and price in region i, and let F2= F2(θ) (F2’>0, F2’’>0) => F2 a fct of share of people connected

• Write welfare function F W=S(q1) +λ q1 .p1 - (1+λ)(c1.q1 – F1) +θS(q2) +θλ q2 .p2 - (1+λ)(θc2.q2 – F2 (θ))

• dW/dθ=0 => (3)• Differentiate this to get • This tells us that the optimal size of network shrinks as

fiscal efficiency shrinks!

2 2 2 2 2( ) '( ) ( (1 )1 )F cS q p q q 0

d

d

36

What if limited fiscal efficiency? (2)

• Solutions??? Look at a typical problem• Imagine rural area more costly, i.e. c2 > c1 , =>ideally: p2>p1

• BUT uniform price restriction to “help” the poor rural area

i.e. implies rather than p2 = (1+λ)c2

• However, (from (3)) this reduces network size• With no government-firm transfers, instead of (3) we have

• => there is a trade-off between affordability and access• Hence when 1+λ >μ, need cross-subsidies targeted to

network expansion to increase network expansion

1 22 (1 )

1

cp

c

2 2 2 2 2'( ) )( ( 1)S q q qcpF

37

Summary of consequences of our institutional problems

  Quantity Quality Cost Prices Welfare

Limited Capacity 0/- - ? + -Limited 

Commitment 0/- - + + -Limited 

Accountability - ? + ? -Limited 

Fiscal Space ? - + ? -38

What solutions does theory offer?  Industry Structure Regulatory Structure Contract Structure

Limited Capacity

Vertical disintegrationMore competition (?)

Centralisation Less independence (?) Fewer regulators Contracting out

Lower powered incentives Simpler contracts

Limited Commitment

Vertical integrationLess privatisation (?)

More independence Multiple regulators Pro-industry bias (?)

Lower powered incentives ? Less discretion 

Limited Accountability

Vertical disintegrationMore competition (?)More privatisation (?)

Decentralisation Less independence (?) Multiple regulators      Anti-industry bias

Lower powered incentives ? Less discretion                                     Fewer cross-subsidies

Limited Fiscal Space

More competition (?)                      Alternative  suppliers                            

Independent subsidy body 

More cross-subsidies ?

39

CONCLUSIONS• We know from experience that the real impacts of the

various institutional limitations discussed can be large • Main real problem is that the solutions available are

imperfect and OFTEN contradictory• Moreover, we still have huge gaps in our

understanding of issues – No real serious link between finance and regulation in this

field• Moreover, good solutions for LDCs often need to

follow a different path from that taken in developed countries – Thus insufficient and possibly damaging to advocate

simply for a regulatory framework that is closer to some universal ideal.–…and a lot more work to do on this topic!

40