COST-BENEFITS ANALYSIS AND INCENTIVES
IN INFRASTRUCTURE PLANNING AND EVALUATION:
A RESEARCH AGENDA FOR THE EU COHESION POLICY
Massimo Florio
University of Milan
Department of Economics, Business, and Statistics
Via Conservatorio 7
20122 Milano - Italy
Preliminary draft
5th Milan European Economy Workshop
26-27 May 2006
Please do not quote this version without the author’s consent
Comments are welcome at massimo.florio@unimi.it
Keywords: Cost-Benefits Analysis, Incentives, Infrastructure Planning, Evaluation
Abstract
Europe needs a huge investment effort for (broadly defined) infrastructure in the next decade. A combination of EU
grants, loans by the EIB and the EBRD, and their leverage effect on private capital is going to mobilise a huge amount
of savings. Planners should establish priorities and criteria. Modern growth theory offers a framework for empirical
research at macroeconomic level, but it is not robust enough to be used for actual investment planning. Microeconomic
social accounting, i.e. cost-benefit analysis, despite its limitations, is more reliable as a support to investment planning.
The key message of modern CBA theory is that shadow prices are not proxies of perfect markets outcome, but are
planning signals that solve a (policy–constrained) social planner’s problem. Planners must compute shadow prices,
evaluators should use them for project appraisal, and the two functions should not be confused. In principle this
distinction applies at each planning level, but a consensus decision-set should emerge form this process, using a bottomup approach. In a multi-government setting there are, however, information asymmetries that need to be addressed, and
we have to turn to incentive theory. The paper proposes to move away from the current low-powered incentive EU cofinancing mechanism, essentially an investment cost part-reimbursement scheme, towards a more incentive-based
system. Financial and economic analysis, ex ante and ex post, should be linked to an economic performance bonus for
more socially deserving projects. Examples are given of such mechanisms. Planners, managers and evaluators should be
given appropriate incentives to use CBA as cooperative learning game.
Acknowledgements
This paper is part of a wider project at the Department of Economics, University of Milan. I am very grateful for
research assistance to this on going project to Elena Castignola, Cristina Lira, Virginia Maestri, Laura Seritti; for
editorial and organisation support to Jone Cocco-Ordini; for revising the English of earlier versions of the paper to
Susan Boyle. A special mention is needed for my research associates in the preparation of the CBA Guide for the
European Commission DG Regional Policy: Francois Levarlet, Silvia Maffii, Mario Genco, Silvia Vignetti, Alessandra
Tracogna, and a special mention to Ugo Finzi, who has helped me to learn from his long project analysis experience at
the World Bank and elsewhere. Some of them have kindly commented on earlier drafts. The usual disclaimer applies,
and in no way what I have written here should be seen as involving the European Commission or any other institution. I
am grateful for financial support in the organisation of the 5 th Milan European Economy Workshop to the European
Commission – Jean Monnet action, and to the Department of Economics, University of Milan.
2
Table of contents
Introduction
1.
Introduction: Overview of the research project
2.
The EU infrastructure agenda
2.1 The regional policy framework
2.3 Grant mechanisms: The Structural Funds and the Cohesion Fund
2.4 Loan mechanisms: The European Investment Bank and the EBRD
2.5 Infrastructure and growth models
3.
Public investment planning and evaluation: theory
3.2 CBA in a multi-government setting
3.3 The economic environment
3.4 Social planners, objectives, and policy tools
3.5 Planning and evaluating
3.6 Shadow pricing rules and their use
4.
Information and incentive issues
4.1 Beyond the planning black box
4.2 The principal-agent framework: an example
4.3 Linking evaluation and co-financing contracts
4.4 Using project return benchmarks as incentives
4.5 On the EU co-financing mechanism
5.
Concluding remarks
Appendix 1.
Theory of CBA in the DS framework
Appendix 2.
An evaluation model in the LT framework
References
3
4
Introduction
Some years ago, in 1994 the European Commission, Directorate General for Regional Policy,
asked me to write a Guide of Cost Benefit Analysis in the framework of the EU Structural Funds
(“the CBA Guide”). The objective of the Guide was to offer a concise reference text for the country
desk officers in Brussels, and guidelines for their national counterparts in the public sector and in
the consulting agencies. The latter are regularly involved in the preparation of project appraisals to
be co-financed by the EU Structural funds, while the former are responsible for reviewing such
appraisals. The CBA Guide had to be simple, practical and straightforward. Two subsequent
editions followed, respectively in 1997 and in 2002. The success of the Guide has been
considerable1. There are probably two reasons for this revival of CBA in Europe, after some years
of less interest elsewhere. First, there is a wide perception that infrastructure are going to play an
important role in European integration. Second, the European Union, through its Cohesion policy
and other frameworks, is a key player in the planning of infrastructure, along with national public
governments and private investors. Substantial leverage effects are expected through public private
partnerships, loan finance (including from the European Investment Bank and the European Bank
for Reconstruction and Development), and other funding mechanisms.
In this context, I think there is some scope for a contribution on how social CBA can be
helpful to infrastructure planning and evaluation in this new European context. I suggest that the
key issues are how to offer flexible but robust methods, and how to generate the right signals and
incentives in evaluation activities. A new generation of public sector officers and consultants,
particularly in the new EU member states, is going to manage thousands of infrastructure project
dossiers in coming years. I have had the opportunity of meeting a fair sample of these dedicated
people over the last few years at evaluation seminars and conferences in Bratislava, Brussels,
Budapest, Edinburgh, Milan, Prague, Rome, Seville, Warsaw, or in candidate countries such as
Croatia and Romania. There are frequently asked questions in two directions: planning and
evaluation methods, and how to design incentives for their implementation.
1
The 2002 edition is downloadable at
http://europa.eu.int/comm/regional_policy/sources/docgener/guides/guide_en.htm (several languages are available in
pdf version, including Italian, German, French, Spanish, Portuguese, Greek, Slovenian, Serbian, Polish) or at
www.csildevelopment.com. Translation in other languages are available in the Internet.
5
Examples of the many questions about methods are: Should we use a unique European social
discount rate under the EU structural funds, or rather several national ones? What is the appropriate
shadow price of passengers’ time-savings for the Trans-European Networks in transport? To what
extent should distribution issues be considered in the context of regional policy?
Examples of questions about evaluation incentives are: How should the European
Commission design financial incentives to stimulate evaluation efforts? Why evaluators so often
neglect risk analysis? How can planners and evaluators learn lessons from ex-post project
appraisal?
Both kind of questions are important. Good methods without the right incentives to use them
will be set aside and the whole evaluation process will not deliver convincing results2.
The rest of the paper is in five sections: first, I present an overview of the on-going research
project; second, there is a discussion on infrastructures and growth; third, I present general CBA
framework in a multi-government setting; fourth, I discuss incentive issues in this context. Some
conclusive remarks are given in the last section.
2
In this paper I will use project evaluation and CBA as synonymous, while in fact the former is a wider concept than
the latter, because it includes other aspects.
6
1.
Overview of the research project
Twenty years after their seminal contribution to CBA techniques, Ian Little and James
Mirrlees, looked with a critical eye at the World Bank experience:
“Good project appraisal is done by people with their own incentives, within organizations that
wittingly or not set these incentives. Both environments of project appraisal, the intellectual and
political-organizational, are keys to the quality of selection overall. This needs to be most
seriously considered by those who manage and create this environment” (Little and Mirrlees,
1994, p.225).
Economists are often inclined to think that because analytical tools exist, they will be used in
practice. This is a bit naive. Project evaluation based on CBA, even in its most simplified versions,
requires research efforts, is time consuming, and in some cases may be controversial. Thus,
decision-makers need to reward good analyses and to punish the sloppy ones. To do that, they
should have an evaluation and planning strategy, not just a set of accounting rules. They need to use
cumulative information on ex-ante and ex-post project returns to establish benchmarks and
yardsticks, and to learn.
This incentive-based approach to project evaluation is feasible and useful, and social CBA, in
spite of its limitation, is the best common language available to us for framing these mechanisms.
This is true in infrastructure–based industries in developing and developed countries as well.
Although many distortions have been addressed in the last few years, observable prices of energy,
water or of transport services are far from their social opportunity costs in many countries. Some
policy reforms aim at “getting price right”, but first, there are often constraints to their
implementation, and second, in some cases they are mistaken. For example, it is not always
desirable that tariffs should be cost-reflective, when income distribution is far from being optimal.
Privatization, liberalization and regulation can be good or bad. They may create private rents and
increase the wedge between market prices and social opportunity costs. Hence, very often, observed
prices cannot be trusted for investment decisions involving public funds. This is probably true
everywhere, albeit with country or region-specific market and policy failures, and with a complex
interplay of central and local governments.
The EU context poses new, interesting questions on infrastructure planning and evaluation in
a multi-government setting. The European Commission and national planners should provide better
7
designed incentives to project evaluation beyond the dissemination of basic principles, as it has
done by establishing guidelines and organising training seminars and evaluation conferences. They
should agree harmonised rules on the calculation of some key shadow prices and performance
indicators, and use them to steer the process. Moreover, they should use the information of ex-ante
and ex-post analyses as an incentive mechanism for generating good projects evaluation and
management. Only a serious dialogue among the many players in the EU planning process can offer
appropriate answers to some of the open questions discussed here. To a large extent, this is an
unprecedented experiment in economic and political integration that deserves attention worldwide.
A tentative list of open research topics in this context – only to a limited extent covered by
this paper - include the following ones:
a) Infrastructure returns: in macro-econometric modelling and in CBA
There is a growing empirical literature on infrastructure and growth in Europe. This literature
is often rather inconclusive because of the limitations of aggregate modelling of public investment.
There is indeed some evidence that sustained infrastructure investment in Europe in the next twenty
years can be an important ingredient for growth recovery. Looking at regional level (NUTS II) data,
infrastructure investment and GDP growth are correlated, but only under certain conditions3. After
all, having a new highway is good only if somebody is going to use it, if contractors do not earn
excessive rents, if users have substantial net benefits and externalities are taken into account. In
principle the micro social rates of returns as estimated in a CBA framework for this type of
investments should be consistent with the equivalent returns in macro-econometric modelling. This
double-check micro-macro analysis has not been yet attempted, and it would be interesting to
perform some explorative work in this area to establish, if possible, broad investment priorities.
b) Integrating CBA and incentive theory
Turning to the microeconomics of public investment, we need an analytical framework for
planning and evaluation. Drèze and Stern (1987, 1990) offer a good starting point, a general theory
that includes a number of earlier works as special cases. Laffont and Tirole (1993) and several other
contributions in incentive theory, including Laffont (2005) offer a framework for the analysis of
information and incentive structure. There are two research avenues to integrate these frameworks,
discussed respectively in sections 3 and 4. First, one should consider a situation where CBA is
decentralized in a multi-government setting, with a supra-national social planner who acts as co-
8
financier. Second, one has to look at the information and incentive issue in this context. The shadow
price definition relevant here is the social opportunity cost of a change in the world. The DS
framework needs to be re-stated in a multi-country setting and under incomplete and asymmetric
information. The economy to be considered includes different types of agents: individuals
(consumers, workers, taxpayers), managers of private and public firms, regional governments,
national governments, and one supra-national planner. They have preferences and react to signals,
such as prices and rations. Some of these signals are under the control of the regional government,
other under the control of other levels of government, other are exogenous for all planners. Each
planner may compute its set of shadow prices, based on its own preferences and constraints, and use
them to evaluate projects. The interplay of planners can be modelled in different ways, including
games, vector optimization, or a bottom-up lexicographic approach. The latter means that when cofunding or licensing decisions involve different planners, the projects selected are those mutually
compatible, i.e. they pass simultaneously a CBA test. This is the ‘consensus decision set’ which
can, however, be an empty set. Under asymmetric information in any case one government or the
implementing firm may have an incentive to manipulate project evaluations to get co-financing
from the supra-national planner. Moreover, evaluation is costly. Hence, there should be a
mechanism to stimulate evaluation efforts and to elicit true information. One of many possible
mechanisms uses cumulative information, project return benchmarks, and ex-post performance
incentives. Over time, fund allocation will reward (punish) good (bad) project evaluators, planners
and managers.
c) Project evaluation in the real world
To what extent the above abstract framework fits in real world project evaluation? There are
two ways of looking at this issue. First, one can consider a number of official evaluation guidelines
that are relevant for infrastructure planning. For example, the CBA Guide adopted by DG Regional
Policy can be compared with some national traditions of public sector project evaluation,
particularly in the UK (the Green Book), France (Le Plan), and in some other EU countries. A
comparison with the official CBA guidelines by the US Federal government (Circular A-94), some
federal agencies and the World Bank may help to understand where we are in Europe. This
comparison with what is done on the other side of the Atlantic is particularly relevant for the EU
because of the quasi-federal nature of EU capital grants for infrastructure project financing. Second,
project appraisal guidelines are one thing, their implementation is an other thing. Frequently
observed errors and omission need to be openly discussed, including optimism bias, unrealistic
3
See section 2.5 below for a preliminary discussion.
9
schedules, inconsistent shadow pricing, and neglect of risk. The real world seems to be two steps
away from theory: written guidelines need to simplify a lot, and are of varying quality and
consistency with theory. In turn, practice on the ground is often far from the official guidelines.
Understanding why practice and theory are so divorced in the CBA arena will help to design better
incentive mechanisms.
d) How many social discount rates?
A standard topic in CBA, that needs further inquiry in a multi-government setting, is the
social discount rate. Different theories are available, but an approach based on the social time
preference, involving forecasts of the consumption growth rate and the elasticity of marginal utility
of income is often deemed to be the most appropriate by economists 4. Social discount rates at
national level based on this approach should be estimated and compared with the SDRs officially
adopted by the Member States5. While in principle the SDR is a national shadow pricing parameter,
in fact probably only two rates are worth using: one for the Cohesion countries, and a lower one for
the remaining EU Member States6. The main rationale for this double regime is the difference in
growth rates across the two groups. The alternative of two or more macro-regional SDRs should be
considered along with its relationship with country-based estimates.
e) Shadow wages and the social value of job creation
The social value of reducing unemployment through public investment spending in the EU is
an important topic in the context of planning the Structural Funds. After considering different
theoretical models, and data about labour conditions at a regional level, it seems that, unlike the
social discount rate, the shadow wage rate is intrinsically region-specific, as it reflects highly
idiosyncratic labour market conditions. A range of wage conversion factors can be estimated for
group of regions, based on three types of unemployment: hidden unemployment in the informal
sector, Keynesian unemployment, frictional unemployment7.
f)
Some critical shadow prices
The shadow prices of labour and capital are two typical planning parameters. There are three
other sets of shadow prices that need to be discussed in the EU context, looking particularly at the
output of revenue generating infrastructures, such as toll motorways, railroads, energy production
4
See Evans and Setzer (2004), Kula (2002)
See Spackman (2004)
6
See Florio (2006)
7
see Florio (2006)
5
10
and distribution, water, etc. These include the value of time, of life saved, environmental benefits 8.
While tariff policy is the responsibility of national regulators, the EC should use shadow pricing as
a way to reveal the trade-offs in infrastructure investments policies. Shadow-pricing rules should be
harmonised across countries, even if key-parameters for the evaluation of intangibles and
externalities should be region-specific. Transboundary effects and ethical concerns may however
suggest to adopt some European benchmark values. A range of values should be given to illustrate
the variability of shadow prices across the UE, and to dismiss the simplistic view, de facto adopted
by some national governments that shadow prices are important only in less developed countries
where actual prices are supposed to be the most distorted.
g) Welfare distribution matters in project analysis
There are contrasting views on this topic in CBA literature and the question arises whether
one should regard EU Cohesion Policy as re-distributive in the conventional sense. My view is that
while there is a solidarity element in it, the answer is negative. There is little evidence in the EU
legislation that transfers from rich to poor countries and regions are a substitute for transfers from
rich to poor households. Cohesion policy is best depicted as the territorial dimension of European
integration. However, this is not an argument for ignoring the distribution impact of projects. If
projects have adverse welfare effects on the poor they can be socially unaffordable. Moreover, tariff
reforms in the new Member States, where many public services used to go priced below costs, are
linked to EU co-financed infrastructures investments, and the Commission faces trade-offs between
projects with higher tariffs and low need for EU grants and the other way round. Approaches to
including social affordability in the evaluation need to be discussed and examples of regional and
social welfare weights given9.
h) Financial mechanisms
One weak point of standard CBA approaches is that they often overlook the relationship
between financial and economic analysis. The CBA Guide is careful in suggesting two different
performance indicators, an internal rate of return on total costs (FRR/c), and a return on capital
(FRR/k - e.g. private equity in public-private partnerships, or national capital before the EU grant).
Existing data on ex-ante financial returns on large samples of projects need to be evaluated10. This
analysis is instrumental to a critical discussion of the incentive structure of the current co-financing
mechanism under the Structural Funds. It is apparent that some grant mechanisms are cost
8
9
See Atkinson (2006), De Rus (2006), Scapecchi (2006).
See for a discussion of these issues Kula (2002).
11
reimbursement schemes, that offer inadequate efficiency incentives for project appraisal, design and
implementation. Alternative approaches can be suggested, including the integration of financial and
social CBA in the granting schemes, as proposed in section 5.
i)
Returns, risks, and incentive design
Finally, it seems useful to link the discussion of the financial and economic returns of an
infrastructure as calculated by CBA, to a discussion of risk and incentive. First, the key concepts
about the information structure of project evaluation need to be examined. Data on the expected
financial and economic returns in large samples of projects approved by the EC can be used to
reveal the origins of observed wide variability in returns. Ex-ante project returns are generated by a
process where there is some uncertainty and asymmetric information between the European
Commission, and the project proponent. This context needs to be analysed in the framework of
incentive theory and development economics. Mechanisms to generate better evaluation and
planning of EU co-financed infrastructures can be suggested. These revolve around the proposal
that information on ex-ante and ex-post project returns be accumulated within the institutions, and
be used to establish benchmarks and incentives for best performers.
The remaining of the paper, covers, in a very preliminary way, the first two and the last two
topics ((a),(b), and (g), (h)). Working papers related to some of the remaining topics will appear at a
later stage, in 2006-2007.
10
See also on this Florio (1999), Florio and Vignetti (2004, 2005)
12
2.
The EU infrastructure agenda
2.1
Overview
Over the last two decades, the Structural Funds and the Cohesion Fund have co-financed
through grants a very large number of projects in the Member States of the European Union. These
include mainly railways, roads, ports and airports, water distribution and treatment, solid waste
management, but also productive investments, science parks, museums, and many others. Other
sources of infrastructure finance include grants under the Trans-European Networks in transport and
energy, and loans by the European Investment Bank (EIB), or by the European Bank for
Reconstruction and Development (EBRD).
In the coming years the EU institutions, national governments, regional managing authorities,
public and private companies will all face challenging infrastructure needs. In 2007-2013 the EU
Funds will contribute to the infrastructure plans of 27 countries, including ten new members
(mostly former transition economies) and two new candidates to accession (Romania and Bulgaria).
ISPA funds will assist Croatia and other accession candidates. The EU seven-years budget
supporting this effort will draw from a provision of over EUR 300 billion for Cohesion policy. Tab.
2.1 shows the Cohesion Policy Budget, eligibility, priorities and allocations.
A substantial part of the funds is going to be allocated to infrastructure projects, in regions
lagging behind in the endowment of basic stock of capital compared to the rest of the EU.
Moreover, there will be a leverage effect of the EU funds on public and private finance, because in
many cases Brussels will contribute only a part of the cost, and the rest of capital expenditure must
be matched by other sources of finance. Tab. 2.2 gives some figures on the leverage effect.
This paper focuses particularly on infrastructure that support services of general economic
interest in sectors such as transport, electricity and gas, telecommunications, environmental
services, etc. However, most of the issues that I consider are relevant for social infrastructure as
well, such as hospital and educational buildings, and for support to industry and R&D ventures.
13
Tab. 2.1. COHESION POLICY 2007–13 (EUR 336.1 billion)
Programmes and
instruments
Eligibility
Priorities
81.7 % (EUR 251.330 billion)
including the special programme for
the outermost regions
Convergence objective
 Innovation
 Environment
 Accessibility
 Human Resources
Statistical effect: regions with per  Infrastructures
capita GDP < 75 % of EU-15 and >  Administrative
Capacity
75 % of EU-25
 Transport networks
Member States with per capita GNI (TEN-T)
< 90 % of Community average
 Sustainable transport
 Environment
 Renewable energy
Regions with per capita GDP < 75
% of EU-25 average
National and regional
programmes (ERDF, ESF)
Cohesion Fund
Allocations
Allocations
Regional competitiveness
and employment
objective
70.5 % = EUR 177.29 billion
5% = EUR 12.52 billion
24.5 % = EUR 61.518 billion
15.8 % (EUR 48.789 billion)
The Member States propose a list
of regions (NUTS1 or NUTS2)
 Innovation
78.7% = 38.404 billion
 Environment/risk
prevention
Regional programmes(ERDF) ‘Phasing in’ regions covered by
and national programmes (ESF) Objective 1 between 2000 and2006  Accessibility
and not covered by the employment  European employment 21.3 % = EUR 10.385 billion
strategy convergence objective
 Strategy
European territorial
cooperation objective
2.4 % (EUR 7.5 billion)
35.61 % cross-border cooperation
Cross-border and transnational
programmes and networks
Border regions and large
(ERDF)
transnational cooperation regions
 Innovation
 Environment/risk
prevention
 Accessibility
 Culture, education
12.12 % European neighbourhood and
partnership instrument
47.73 % transnational cooperation
4.54 % networks
Total
307,5
Source: European Union Regional Policy, Factsheet 2004: “Cohesion Policy: the 2007 Watershed. Legislative
proposal by the European Commission for the reform of the Cohesion Policy (2007-2013 period)”, Council of the
European Union: “Financial Perspective 2007-2013”, p. 8. Brussels, 19/12/2005.
Tab. 2.2. - Leverage effect of Structural Funds on public and private expenditure under Objective 1, 1994-1999
and 2000-2006 (EUR)
1994-1999*
2000-2006
National public funds
Private funds
National public funds
per euro of SF
per euro of SF
per euro of SF
0.77
1.18
1.02
BE
0.37
1.53
0.58
DE
0.52
0.28
0.50
EL
0.51
:
0.52
ES
0.54
0.23
0.88
FR
0.43
0.34
0.76
IE
1.40
:
0.89
IT
2.49
1.42
2.15
NL
1.59
3.79
0.33
AT
0.42
0.30
0.60
PT
0.53
0.24
0.85
UK
0.62
0.36
0.63
Total EU11
Notes: * based on actual expenditure 1994-2000
ES, IT: for 1994-1999, national public funds include private funds; EU11: excluding FI, SE
Private funds
per euro of SF
1.43
0.02
0.48
0.04
0.33
0.25
0.45
0.55
1.76
0.46
0.43
0.29
Source: Third report on economic and social cohesion: A new partnership for cohesion convergence competitiveness
cooperation. Statistical Annex to Part 4: Impact and added value of structural policies, p.180, EC, 2004.
14
Some authors have taken a highly critical attitude about the impact of these EU funding
mechanisms, and have even proposed a discontinuation of the Structural Funds. The Sapir Report
(Sapir et al, 2004) has proposed a wide reform, which in fact amounts to concentrating available EU
resources on the new Member States, and to entirely delegating the project planning to them. While
the Sapir Report offers plenty of good suggestions, I am not convinced of re-nationalization of
regional policy, and I would rather advocate a stronger role for the European Commission in this
area (Florio, 2005). The EC is in an unique position to capitalise infrastructure knowledge across
countries and regions, and this learning mechanism has an intrinsic value, that will be entirely lost
by full re-nationalisation of planning and evaluation. A consistent use of social CBA should be seen
as the common language for this learning mechanism.
Moreover, there is an interplay between regional policy and other crucial EU integration
policies, for example the creation of European wide energy markets, of trans-boundary multimodal
networks, and of standards in environmental protection. In fact Structural Funds overlap with these
broader policy frameworks, because infrastructure for services of general economic interest should
be a component of the wider European project11.
I suggest that in future the EC should be given a stronger financial basis and power to act as a
quasi-federal body for infrastructure planning, through a system of executive agencies. This
proposal may be at variance with the current political climate in several EU member states, where
there is some disenchantment with the European project, but I argue that infrastructure and related
services, planning, evaluation, and co-financing are good examples of the need for an active role of
the Commission. Moreover, I think that this wide-scale experiment has importance for the rest of
the world, that from EU successes and failures, can learn how to manage complex multi-lateral
investment plans.
In the rest of this section I briefly present some institutional features of EU infrastructure
funding: grants by the Structural Funds and the Cohesion Fund, which are the key-mechanisms
managed by the European Commission; and loans by the EIB and the EBRD, two financial
institutions co-sponsored by the EU. I turn then to the macroeconomic planning approach.
11
This peculiar EU policy setting contrasts sharply with some limitations of international aid to developing countries
where the supra-national policy context is often weak. International lenders, such as the World Bank or e.g. the InterAmerican Development Bank in Latin America, do not have the instruments to fully exploit potential externalities of a
large-scale regional infrastructure strategy. In the US there was an important tradition in infrastructure planning, but it
seems that social CBA is no more consistently applied in the appraisal of public sector projects.
15
2.2
Grant mechanisms: The Structural Funds and the Cohesion Fund
The EU Structural funds are financial instruments that offer Community assistance, in the
form of capital grants, to different kinds of regional programmes and project. Table 2.3 shows some
figures from the previous planning period (2000-2006).
In the framework of the 2007-2013 Cohesion Policy there are three main objectives. The first
one, and by far the most important in terms of funds available under the Cohesion Policy (around
82%), is the objective of supporting the convergence of sustainable economic growth in the regions
lagging behind. Most of these regions are located in the new Member States, but there are many
relatively under-developed regions in some rich countries in the former EU-15, particularly in Italy
(the Mezzogiorno), in Germany (the Eastern Laender of the former DDR), in Spain, Greece,
Portugal, in the overseas French islands, and elsewhere. A second objective is to increase the
competitiveness and employment outlook in some of the remaining regions. Many of them, while
located in the core areas of Europe, face high unemployment and relatively modest growth. Third,
there is an objective of territorial co-operation that is of some relevance for regions facing transboundary problems and in some specific geographic conditions.
EU assistance to achieve these objectives revolves around a small number of financial
instruments, each with a set of operating rules, eligibility conditions, co-financing rates, etc. The
most important of these funds is the European Regional Development Fund (ERDF).
In the Convergence regions, defined as those where GDP per capita is below the threshold of
75% of the EU average in terms of purchasing power standard Euro, the ERDF has a very wide
range of possible intervention areas12.
Eligible investment projects in the Competitiveness regions are more focussed on three
priorities: innovation and the knowledge economy, environment and risk protection, and
accessibility (transport and TLC). Under the Territorial Cooperation objective, the priorities are
cross-border, joint development programmes, trans-national cooperation in infrastructure for
accessibility and the environment, and networking of regions. There are also specific provisions for
urban and rural areas, and for some areas with particular natural handicaps.
12
These include inter alia: research and development, innovation and entrepreneurship, development of business
clusters, support to SMEs; information society projects, including adoption of ICTs by small and medium enterprises;
environmental projects, including water, waste management, air quality, rehabilitation of contaminated land, pollutionpreventing technologies; natural and technological risk prevention; promotion of sustainable tourism and enhancement
of the cultural heritage; transport investment (rail, highways, ports, airports), including the trans-European networks and
clean urban transport; energy investment (electricity and gas, etc) including the trans-European networks; education
infrastructures; health infrastructures; direct aid to investment of SMEs for job creation or safeguard of existing
employment.
16
Tab. 2.3.- Use of Structural funds in 2000-06 period by Objective and field of intervention (%)
Objective 1
Objective 2
Objective 3
1
Productive Environment
34.96
55.83
0.55
1.0
Productive Environment
0.06
0.06
0
1.1
Agriculture
14.55
0.22
18.95
1.2
Forestry
3.35
0.08
19.47
1.3
Promoting the adaptation and the devepolment of rural areas
18.58
4.14
0.53
1.4
Fisheries
5.50
0.16
0.53
1.5
Assisting large business organizations
7.76
4.21
4.17
1.6
Assisting SMEs and large business organizations
27.30
57.51
25.22
1.7
Tourism
8.80
15.81
20
1.8
Research, technological Development and Innovation (RTDI)
14.09
17.80
11.14
2
Human resources
23.27
10.56
97.01
2.0
Human resources
0.28
5.17
0.21
2.1
Labour market policy
30.29
17.29
29.98
2.2
Social inclusion
13.92
18.78
21.96
2.3
Developing educational and vocational training not linked to a specific sector
30.29
19.93
20.08
2.4
19.83
33.49
20.94
2.5
Workforce flexibility, enterpreneurial activity, innovation, information and
communication techologies
Positive labour market actions for women
5.39
5.34
6.83
3
Basic Infrastructure
39.54
28.49
0.35
3.0
Basic Infrastructure
0.00
3.08
0
3.1
Transport infrastructure
48.14
20.41
0
3.2
Telecommunications infrastructure and information society
9.17
11.31
91.52
3.3
Energy infrastructure
2.50
3.19
0
3.4
Environmental infrastructure
16.36
14.08
0
3.5
Planning and rehabilitation
14.33
44.20
0
3.6
Social and public health infrastructure
9.50
3.74
8.48
4
Miscellaneous
2.23
5.12
2.09
4.0
Miscellaneous
2.23
3.11
0
4.1
Technical assistance and innovative actions (ERDF, ESF, EAGGF,FIFG)
88.46
42.66
100
4.9
Miscellaneous
3.59
54.23
0
Source: 16th Annual Report on the implementation of the Structural Funds 2004. Technical annexes,
Pp .183-193. Brussels, 28/10/2005
Project selection and evaluation within this very broad (perhaps too broad) framework is
normally the sole responsibility of the national authorities. However for very large projects (with a
total investment cost of more than EUR 50 million, or 25 for environmental projects), the EC
requires Member States to submit a cost-benefit analysis and then takes a specific co-financing
decision. Ceilings for EU- co-financing are different according the region and the fund. Moreover,
ERDF finance, in form of a grant, can be combined with loans by the EIB, see below, and with
other sources of finance.
17
Fig. 2. 1 – Structural Fund expenditure on Transport under Obj.1 2000-2006
35
30
25
20
Serie1
15
10
5
0
Roads
Rail
Motorways
Urban
transport
Ports
Multimodal
transport
Airports
Other
Waterways
Intelligent
Transport
Systems
Source: Third report on economic and social cohesion: A new partnership for cohesion convergence competitiveness
cooperation. Statistical Annex to “Part 4: Impact and added value of structural policies”, pg.183. 02/2004.
The Cba legal base for the cohesion policy
CBA is explicitly required by current Structural Funds Regulation (Reg. 1260/99), Cohesion
Fund (Reg. 1264/99; Reg. 1265/99) and ISPA (Reg. 1267/99), for projects with a budget,
respectively, of more than EUR 50, 10 and 5 million. Regulations state that Member States have the
responsibility for prior appraisal, while the Commission must verify that information provided in
the appraisal are exhaustive so to allow project selection and determination of the EU co-financing
rate. Community regulations indicate which information must be contained in the application form
for the purposes of an effective evaluation on the part of the Commission. Article 26 of reg.
1260/99, for co-financing of major projects, asks for:
“a cost-benefit analysis, - a risk analysis, - an evaluation of the environmental impact (and the
application of the Polluter Pays Principle), - the assessment of impact on equal opportunities and on
employment”.
In the words of the CF Regulation: Art 1(2) Reg 1265/1999:
“Beneficiary Member States shall provide all necessary information (…) including the results of
feasibility studies and ex ante appraisals. (...) Member States shall also provide, (…) where appropriate,
an indication of the possible alternatives that were not chosen.”
18
Moreover it specifies that the project dossier must include
“(…) the actual starting date of the project,- the way in which it will be managed once finished, confirmation, if appropriate, of the financial forecasts, especially as regards the operating costs and
expected revenue, - confirmation of the socio-economic forecasts, in particular the expected costs and
benefits, - an indication of the environmental protection measures taken, and their cost, including
compliance with the polluter-pays principle”
Project feasibility does include engineering aspects, and in many cases, it also concerns
marketing, management, analysis of the implementation, etc. There are often various project
options in order to achieve a socio-economic objective. Regulation for the Cohesion fund and the
ISPA, in addition to stating that the proposals for co-financing must contain a cost-benefit analysis,
also provide some indications of the criteria to be applied in order to ensure the quality of the
evaluation:
“financial plan that includes, wherever possible, information about the economic viability of the
project”
(see art. 10 (4), reg. 1164/94) and, in the case of environmental projects:
“a cost-benefit analysis supplemented by other evaluation methods, possibly of a quantitative
nature such as a multicriteria analysis and the consideration of the Polluter Pays Principle (see art. 10
(5), Reg. 1164/94 and the Council’s amendments)”.
ISPA and Cohesion Fund Guidelines and Application forms are more explicit in the
methodological requirements of the project dossiers submitted for co-financing. For example, as
regards the financial analysis the CF Guidelines ask for:
“All applications must be accompanied by a financial analysis, attached in annex, indicating: description of methodology (with particular mention to the assumptions made); - economic life of
project;
- capital costs; - the revenues generated (if any); - the operating and maintenance costs over its
lifetime: - revenues generated over its lifetime; - discounted cash flow (DCF) analysis; - sensitivity
analysis;”
And moreover:
“All applications must be accompanied by a cost-benefit analysis, or other quantified
economic analysis, such as cost effectiveness analysis or multi-criteria analysis. In the case of costbenefit analysis, it should include the following: - description of methodology; - alternative options
considered; - direct and indirect costs and benefits in construction stage; - direct and indirect costs
and benefits in operational stage; - key assumptions made in valuing costs and benefits; - assessment
of costs and benefits which cannot be fully quantified or valued; - main beneficiaries of the project(s)
and anticipated rate of utilisation; - results of analysis expressed in terms of Internal Rate of Return,
Net Present Value or benefit-cost ratios; - assessment of risk and uncertainties (estimated effect of
changes in main parameters); - conclusions”13.
13
EC, Guide to the cohesion fund 2000-2006, February 2000.
19
The new legal base for CBA of investment project funded within the cohesion policy is given
by the draft regulation of Structural Funds and Cohesion Fund. Art 39 of the Presidency
compromise of 7 Apr. 2006 states that:
“The Member State or the managing authority shall provide the Commission with the following
information on major projects:
a) information on the body to be responsible for implementation; b) information on the nature of
the investment and a description of it, its financial volume and location; c)
feasibility studies; d)
the
results
of
the
a timetable for implementing the project and, where the implementation
period for the operation concerned is expected to be longer than the programming period, the phases for
which Community co-financing is requested during the 2007-2013 programming period; e) a cost-benefit
analysis, including a risk assessment and the foreseeable impact on the sector concerned and on the socioeconomic situation of the Member State and/or the region and, when possible, of the other regions of the
Community; f)
an analysis of the environmental impact; g) a justification for the public contribution;
h) the financing plan showing the total planned financial resources and the planned contribution from the
Funds, the EIB, the EIF and all other sources of Community financing, including the indicative annual
plan of the financial contribution from the ERDF or the Cohesion Fund for the major project. The
Commission shall adopt indicative guidance on the methodology to be used in carrying out the cost
benefit analysis foreseen in (e) above in accordance with the procedure referred to in Article 104(2)”.
While the ERDF is in a broad sense targeted at infrastructure and productive investment, the
European Social Fund is mainly concerned with human capital, including support to vocational
training and education programmes of different nature, public or private.
Lastly, the Cohesion Fund was established in 1993 under the Maastricht Treaty to promote
economic and social cohesion and solidarity between EU Member States. It funds projects in the
field of the environment and Trans-European transport infrastructure networks. The rationale for
establishing the CF was that the least prosperous Member States should be helped to invest heavily
to strengthen their growth potential. Member States eligible for CF assistance are those whose per
capita gross domestic product (GDP) measured in purchasing power parity is less than 90% of the
EU average. These countries originally were Greece, Portugal, Ireland and Spain. As from 1 May
2004, the new EU Member States are all eligible, and they will receive around EUR 8.5 billion until
2006. The total CF budget for 2000-06 amounts to EUR 18 billion (1999 prices) for EU 15. It went
up from EUR 1.5 billion in 1993 to more than EUR 2.6 billion in 1999 (1992 prices), see figures 2.1
and 2.2.
20
Fig.2.2. – 2004 CF commitment appropriations by Member States
100
90
80
70
60
50
40
30
20
10
0
Spain
Greece
Portugal
Environment (%)
Cyprus
Czech
Republic
Estonia
Transports (%)
Hungary
Latvia
Lithuania
Malta
Poland
Slovakia
Slovenia
Percentage of the total CF appropriations (%)
Source: Annual Report on the Cohesion Fund 2004, Technical Annex. p.31
As for the 2007-2013, the CF is one of the three funds out of the previous six that remain as
instruments for the convergence objectives. The EC proposal for the allocation to the CF in 20072013 is EUR 61.5 billion, or 24,5% of the Cohesion Policy budget. Eligible countries will again be
those with GDP below the 90% of the EU average. This will include Greece, Portugal, Spain, the
current ten new members plus Romania and Bulgaria if they are going to join the EU in 2007 as
forecasted by the accession treaties. Eligible investment projects will include Trans-European
transport networks, sustainable transport, environment, and renewable energy.
21
Fig.2.3. – Breakdown of CF allocations for the acceding countries 2004-06
4500
4000
3500
3000
2500
Mid-range allocations (€ million – 2004 prices)
53.94*
2000
1500
1000
500
0
Czech
Republic
Estonia
Hungary
Latvia
Lithuania
Malta
Poland
Slovakia
Slovenia
Source: Annual Report on Cohesion Fund 2004. Annex, pg.16. Brussels, (2005)
2.3.
Loan mechanisms: the EIB and the EBRD
Two European financial institutions play an active role in providing infrastructure finance,
mainly in the form of long-term loans and other financial arrangements: The EIB and the EBRD.
The EIB is an international financial institution, created in 1958 by the six founding countries
of the European Economic Community and established by the Treaty of Rome. The mission of the
bank is to further the policy objectives of the European Union by providing long term finance for
capital projects. The first operational priority for the Bank 14 is the promotion of economic and
social cohesion within the European Union, by reducing the imbalances existing among regions or
social groups.
The EIB operates by raising significant volumes of capital on the world markets, particularly
by bond issues. These enjoy top credit ratings by independent agencies, so they can pay low returns,
and finance EIB customers at the most favourable rate and maturity term. Such terms are set, from
time to time, on the basis of the specific project requirements.
14
See: Article 267 of the EIB’s Statute.
22
Tab.2.4 – EIB Loans provided within the European Union from 2000 to 2004
Breakdown by country and sector. EUR million
Country
Infrastructure
Total
2 624
Belgium
Czech Republic 3 451
5 005
Denmark
31 638
Germany
237
Estonia
7 055
Greece
27 202
Spain
19 341
France
2 482
Ireland
28 952
Italy
740
Cyprus
218
Latvia
165
Lithuania
529
Luxembourg
2 784
Hungary
25
Malta
2 662
Netherlands
4 383
Austria
6 196
Poland
8 653
Portugal
804
Slovenia
805
Slovakia
3 647
Finland
3 937
Sweden
15505
United
Kingdom
1 240
Other
180 280
Total
Individual
loans
Energy
Communications
Water management Industry
Health
and sundry
Services
Education
infrastructure
Agriculture
Global
loans
1 659
3 016
4 577
70
–
831
592
1 523
2 507
850
710
124
147
688
766
–
95
349
965
435
429
16 899
97
6 845
568
80
587
3 937
5
3 837
3 053
12
1 750
5 697
–
247
3 644
–
424
14 739
140
210
19 220
8 780
2 133
–
12 750
5 371
2 100
275
1 365
2 114
872
1 020
7 982
10 561
1 637
18 419
740
717
4 545
200
753
6 959
55
100
4 426
100
–
2 361
70
68
128
315
845
10 533
–
128
125
529
80
–
80
48
65
135
–
60
184
–
–
130
–
–
–
90
40
–
2 079
–
247
–
1 076
–
273
–
319
–
164
–
705
25
1 724
2 484
5 193
–
82
30
875
556
3 003
770
516
1 256
79
669
430
–
661
475
938
1 899
1 003
6 703
614
585
727
1
–
5 061
568
254
500
45
42
415
–
274
–
–
14
1 950
190
220
3 021
3 587
18
343
318
1962
684
215
1 050
633
950
434
625
350
12 810
2 355
4 012
3 653
1469
1 322
2 695
1 140
122 610
567
14 262
241
56 458
33
21 731
300
19 224
–
10 935
100
57 670*
Source: The EIB Group Statistical Report 2004. Table B
The bank loans do not discriminate between public and private customers, and are of two
types: “global loans”, for small and medium scale projects (from EUR 40,000 to 25 million, see tab.
2.4) that should be forwarded by an eligible-intermediary bank, and “individual loans”, for projects
whose values exceed the threshold of EUR 25 million. The activity of the bank is defined according
to its medium-term policy and its specific operational priorities, which are in turn set by a threeyear-span strategic document, the Corporate Operational Plan (COP). In 2005, total lending for
regional development projects was about EUR 34 billion, nearly the 80% of the bank’s aggregate
lending within the Union15.
Another important European financial actor is EBRD, a multilateral development bank
established in 1991. This bank is owned by 60 countries and two institutions, the EU and the EIB.
15
The EIB Group: Social and economic cohesion, (7/12/2005). Beyond supporting EU regional development, the 20062008 COP provides for the EIB also to deal with: the implementation of the Innovation 2010 Initiative; the development
of Trans-European Networks; the support of EU development and cooperation policies in partner countries;
environmental protection and improvement, including climate change and renewable energy; the support for smallmedium sized enterprises, as well as mid-cap companies of intermediate size and finally the support for human
resources and health - See: http://www.eib.org/about/index.asp?designation=objective
23
Its task is to facilitate the transition process from central planning to market oriented economies, in
Central and South-Eastern Europe and in the successor states of the former Soviet Union.
According to the EBRD charter, at least the 60% of its resources should be addressed to support and
promote private entrepreneurial initiatives. The remaining resources should be addressed to public
sector projects that foster private sector development or to projects supporting the privatisation, the
restructuring of state-owned firms and the improvement of municipal services, see table 2.5.
Tab. 2.5 - Distribution of EBRD Banking operating assets, undrawn commitments and guarantees 2004
(EUR million)
Analysis by country
Operating
assets
Undrawn commitments
and guarantees
Albania
Armenia
Azerbaijan
Belarus
Bosnia and Herzegovina
Bulgaria
Croatia
Czech Republic
Estonia
FYR Macedonia
Georgia
Hungary
Kazakhstan
Kyrgyz Republic
Latvia
Lithuania
Moldova
Poland
Romania
Russian Federation
Serbia and Montenegro
Slovak Republic
Slovenia
Tajikistan
Turkmenistan
Ukraine
Uzbekistan
Regional
Total at 31 December
69.4
35.1
175.3
46.1
126.6
318.0
737.2
427.1
254.1
90.9
56.0
688.3
513.2
59.5
72.1
186.3
72.9
1,226.9
1,137.3
2,192.6
244.5
433.8
185.6
18.2
63.7
539.5
153.3
798.4
10,921.9
106.1
4.1
67.1
11.5
179.2
228.0
204.7
49.3
4.8
152.3
27.3
92.9
214.9
7.2
7.5
51.4
13.0
282.9
430.0
1,317.7
348.2
52.1
34.5
8.0
24.6
329.1
123.4
807.4
5,179.2
7,669.8
3,196.1
56.0
–
–
10,921.9
4,033.5
651.1
–
214.1
280.5
5,179.2
345.1
223.8
763.2
568.7
4,377.6
485.0
1,680.3
326.4
831.7
1,320.1
10,921.9
244.3
187.6
745.2
134.2
1,276.4
629.7
522.0
212.9
67.5
1,159.4
5,179.2
Analysis by instrument
Loans
Share investments
Debt securities
Trade finance guarantees
Other guarantees
Total at 31 December
Analysis by sector
Commerce and tourism
Community and social services
Energy/power generation
Extractive industries
Finance
Local authority services
Manufacturing
Primary industries
Telecommunications
Transport and construction
Total at 31 December
Source: EBRD Annual Report 2004: Annual review and Financial report. May 2005
24
Both the EIB and the EBRD use internal evaluation approaches to their operations. The
performance of EBRD operations are systematically self-evaluated by the project teams, and around
half of them are again evaluated by the Project Evaluation department, which combines project
performance rating methods developed at the World Bank and elsewhere, with an original approach
to the assessment of the ‘transition impact’ of the operations.
In the new member states the EU Structural and Cohesion Funds, EIB and EBRD finance can
be combined, in addition to other national and international capital. The preparation of the EU
2007-2013 programming cycle has seen an intensification of the operational cooperation between
the EIB and the European Commission, in support of Economic and Social Cohesion. Thereby, the
EIB strategy is adapting to the new UE strategic framework, laid down by the Lisbon and
Gothenburg strategies for a sustainable and competitive economy and by the European employment
strategy16.
In October 2005, another policy initiative, JASPERS (Joint Assistance for Preparing Projects
in European Regions), a technical assistance partnership among the Commission, the EIB and the
EBRD, was presented with the purpose of better exploiting synergies and developing new
complementarities between the two institutions, to facilitate the successful implementation of the
Cohesion Policy during the next programming period (2007-2013). JASPERS has been designed to
help beneficiary countries to absorb EU Structural and Cohesion Funds to finance investments over
the next planning period. The aim is to ensure that the Structural and Cohesion funds are invested in
sound and viable projects. It basically consists of a core team of fifty experts who will offer their
technical assistance to Member States and Regions, in order to help them in the preparation and
presentation of quality projects, or in the evaluation of large projects set in the structural and
cohesion funds operational framework. The assistance and advisory activity will be concentrated on
projects relating to road and rail transport, urban transport and environmental protection and
improvement (i.e. water, energy efficiency and renewable energy).
2.4
Infrastructure, growth models and macroeconomic planning
How far this sizeable infrastructure investment effort will contribute to growth in the EU?
How economists would advise decision-makers about investment priorities? In this section I briefly
discuss the macroeconomic perspective in investment planning and its limitations.
16
With regard to the Lisbon strategy, in order to better address the issue of SME competitiveness, in 2000 the EIB and
the European Investment Fund (EIF) joint together to create the EIB Group. In fact, the EIB is supposed to grant
medium and long term global loans for SME projects, while the EIF is expected to focus on venture capital activities.
25
In the neoclassical tradition, for a given population, the saving rate and investment determine
the steady state of GDP per capita (Barro and Sala-i-Martin 2004). In the Solow model, the returns
on capital are decreasing, and growth is a transitory process in response to exogenous shocks,
including new technology, which is freely available. In this framework regional convergence can be
accelerated by transfer of capital from rich to poor regions. In spite of a wide empirical literature,
however, the macroeconomic evidence of a consistently positive correlation between infrastructures
and macroeconomic growth variables, across countries and time periods, is mixed. There is perhaps
a simple theoretical explanation for this: in the long term infrastructure are beneficial to GDP
growth and to other, more comprehensive, welfare measures, only if they are complementary to
firms’ investment or to household consumption.
While private investment is often strongly correlated to output growth, because on average
firms have an incentive to select good projects, most infrastructures are public investments. Public
investment is not constrained by a profitability criterion, thus it can be bad or good, depending on
the orientation and quality of the underlying decision-making and implementation process17.
Moreover, market signals for railways, roads, telecommunications, electricity, gas, water
management, education or health services etc. often do not give the agents the information needed
to achieve socially efficient outcomes. While some think that the problem is self-correcting through
industry liberalisation, the evidence shows that in the EU, as elsewhere in the world, infrastructure
economics is the realm of imperfect markets. This happens because of a combination of market
failures, policy objectives and constraints, including regulatory and distribution issues.
In fact, empirics of public infrastructure and growth is still an open field of research and
debate. In the last twenty years many economists have invested in research on public investment18.
The standard reference was the paper by Aschauer (1989), whose simple idea was to add public
capital stock to a Cobb-Douglas production function, and to proxy total factor productivity by a
constant and a trend variable. After controlling for the business cycle by a rate of capacity
utilization, assuming constant returns on inputs, and taking natural logarithms, the estimated output
elasticity of public capital for the US 1949-85 years was 0,39. This suggested a very high return
indeed for public investment. In fact, while public investment as a share of GDP together with per
capita output increased steadily after the end of World War II, the trend was reversed in the ‘70s
and the ‘80s: public investment declined in the US, and productivity also declined. Perhaps the
former variable caused the latter? Tenths of papers were suddenly produced that tried to confirm or
17
18
Public-private partnership involve crucially public sector decision-making as well.
This section draws from Florio (2000).
26
to disproof Aschauer’s findings. His paper became one of the most cited references in applied
economics literature.
Many years on, it is time to take a cool view of all this research. Sturm (1998) offers a survey
of where we were at the end of the ‘90s.
Sturm identifies five different research strategies:
-
the production function approach (by far the most popular);
-
cost-functions, (this is named “behavioural approach”);
-
VAR studies (increasing in more recent years);
-
cross-country or regional cross-section growth regressions (highly constrained by data
availability);
-
public investment in structural econometric models (perhaps a bit out of fashion, but still
used).
A cursory reading of the results of this tremendous amount of research (further increased in
more recent years) suggests that reported output elasticity of public capital in different countries,
times, estimated in different ways, may lie somewhere between 0,03 and 0,73: a wide range indeed
(and often estimated parameters tend to be not significant). However, an average across authors for
the US is still near 0,30, not very far from Aschauer’s estimate.
The pros and cons of different approaches are discussed by Sturm: the production-functionbased literature is criticized because it disregards issues of data stationariness, functional form
specification, and causality. The behavioural approach is informationally demanding; VAR studies
are still too limited to draw conclusions from them; cross-country studies are not very robust;
structural models are difficult to interpret on sound theoretical grounds19.
Sturm’s overall conclusions are perhaps disappointing: “...public capital probably stimulates
economic growth, but... we are still unsure about the size of this effect.”
19
Sturm’s own contribution to the literature is a set of models he tests on the Netherlands. The comparison of US and
Dutch case-studies with the Aschauer’s original approach is replicated by using first-differences data instead of levels,
in order to eliminate problems arising from variables that are neither stationary nor co-integrated. A different model
uses a symmetric generalized Mc-Fadden cost function and introduces a distinction between a “sheltered” sector ( trade,
banking, other services) and one “exposed” to international competition (manufacturing, agriculture, transport). The
overall result is that a 10% increase of public infrastructure reduced the cost for the private economy by 3%, but the
impact was concentrated in the sheltered sector, while the exposed sector shows statistically insignificant cost impacts
of the infrastructure. This is an interesting result, that may suggest that public infrastructure effects may be quite
different for different sectors, something that is intuitively appealing. A VAR analysis is tried for Dutch 1800-1940
data, including GDP, infrastructures and machinery. The results confirm that infra-structural investment positively
influenced output in the second half of nineteenth Century, in particular transport infrastructures seem to have had longlasting effects compared to other capital expenditures, again a result that is in agreement with our intuition.
27
One may agree that the motivation for public investment should lie less in its aggregate
impact on growth than in a micro analysis of specific infrastructure projects. Perhaps, the main
problem with the measurement of the aggregate impact lies in the poor quality of measurement of
the key variables: the stock of public capital, fixed investments, and Gdp itself (Florio, 2001).
More recently, Calderon and Serven (2004) evaluated the impact of infrastructure on
economic growth with a large panel data including over 100 countries over the period 1960-2000,
with inequality measures and indicators of quantity and quality in addition to standard controls.
They use appropriate estimating techniques for the endogeneity problems and find two robust
results: (a) growth is positively affected by the stock of infrastructures, and (b) income inequality
declines with infrastructures, see also Calderon and Chong (2004).
Turning more specifically to the EU context, structural economic convergence is an ambitious
objective and it depends, at least partially, upon effectiveness and efficiency in the use of
instruments available in the new programming period. The current debate concentrates on the longterm sustainability of the Cohesion Policy of the European Union and its effectiveness with regards
also to the enlargement process. Different opinions on more or less radical reforms of European
interventions have been expressed on these issues20. Most of the debate revolves around the
empirical assessment of the impact of the Structural funds on growth and convergence across
countries and regions. Academic literature does not always agree with the Commission’s optimistic
expectations: see European Commission (2003, 2004).
The European Commission (2004, 146) maintains that there is econometric evidence of
convergence in per capita GDP real growth for 197 (Nuts II) regions between 1980-200121.
Regions with the lowest initial per capita GDP had higher GDP growth (beta convergence, see
Barro and Sala-i-Martin, 2004), and the variance of the log of GDP per capita narrowed as well
(sigma convergence). One question, however, is how to causally relate these findings to the role of
the Structural Funds. According to EC (2004), Objective 1 regions, i.e. those lagging behind, have
converged in per capita income, see table 2.6. In the late ‘90s the remaining regions have however
diverged.
In the 2000-2006 planning period EU transfers to the Objective 1 regions of the EU-15 are
equivalent to EUR 127.5 billion, amounting to 0.9% of GDP in Spain, or 2.4% of GDP in Greece.
20
On the effects of EU Regional Policy on its convergence outcome, see in particular Armstrong (1996), Begg (1999),
Bradley (1995), Fayolle and Lecuyer (2000), Lolos (2001), Pereira (1997). On the impacts of EU Regional policy and
effects of Structural Funds reforms on accession countries see Fayolle (1998), Nicolaides (1999), Swinnen (2001),
Welfens (1999). Empirical evidence, moreover, would suggest contrasting conclusions: together with clear cases of
success, like Ireland (see, for example, Barry, 1999, Barry, Bradley, and Hannan, 2001 and Payne, Mokken and
Stokman 1997), there are less positive experiences, like the Mezzogiorno in Italy. For other country or regional case
studies see e.g. Bristow, Blewitt, (2001) for Wales, Dauce (1998) for Bourgogne, Lolos and Zonzilos (1994) for Greece.
28
In terms of investments, the structural funds contribute on average for example around 9% of fixed
capital expenditures in Portugal, or 7% in the Italian Mezzogiorno. For the new Members States,
EU support in 2007-2013 may reach up to 4% of the country GDP per year. Despite this huge
disbursement of aid, the macroeconomic evidence of causality linking the EU Structural Funds to
GDP growth has been questioned, basically with arguments of potential displacement effects.
Using aggregate production functions, De La Fuente (1996), found that national growth in
Spain would have been higher if support to infrastructure investments had been distributed
according to purely efficiency criteria, instead of regional redistribution to the regions lagging
behind. For a critical view of the EU regional policy, see for example Boldrin and Canova (2003).
They are not confident about the growth impact of enlargement and dismiss the thesis that transfers
from the EU can become growth engines for the new Member States. Such transfers, in their view,
may increase income by an amount equal to the one transferred but there is no evidence that they
have an impact on long-run growth rates. However, estimations made with the EC Hermin model
are supportive of the positive role of Structural Funds (Bradley and Morgenroth, 2003). It is
estimated that real GDP in the ‘90s was 2.2% higher in Greece, 1.4% in Spain, 2.8% in Ireland, and
4.7% in Portugal than it would otherwise have been without the Structural Funds, see Bradley
(2006) for a summary of Hermin-based evaluation.
Some of the more recent empirical contributions on growth in the EU stress the importance of
technological change either in the framework of the “new growth theory” (Kubo, 1995) or in an
evolutionary perspective. According to the new growth theory, technological progress, human
capital and GDP growth are mutually linked and determine an endogenous growth process.
Investment in education sustain human capital, R&D expenditures and learning-by-doing support
technological progress, and both have an impact on growth, which in turn allows for greater
investment in research and education. Public policy in support of these investments can counteract
the diminishing returns of fixed capital (Romer, 1986). In this perspective, while infrastructure is a
generic term, infrastructures that are complementary to education and research are particularly
important in a growth perspective22.
21
Mairate, 2006
The evolutionary or neo-Schumpeterian view stresses the role of innovation in economic growth. Under this view, a
new technology has several dimensions that cannot be studied under the standard competitive markets assumptions. The
conditions for a new technology to emerge have been studied extensively. Differently from the neo-classical model, a
process of catching-up between front-runners and regions lagging behind is not warranted, because it depends upon the
social and institutional framework (Abramovitz, 1986). Consequently, the neo-classical and the endogenous growth
theories offer quite different perspectives in the study of regional convergence in the EU. Empirical tests do not provide
robust results. The most important feature is perhaps that while there is some evidence of convergence of GDP at
22
29
Tab. 2.6 - Regional convergence across EU-15 NUTS2 regions by planning periods and objectives
No. of regions
GDP per head
Beta convergence
R-squared
(% growth rate) per year (%)*
1980-88
All EU-15 regions
Objective 1 regions
Other regions
197
55
142
2.0
1.9
2.0
0.5
0.4
2.1
0.94
0.87
0.92
197
55
142
1.3
1.4
1.2
0.7
3.1
0.8
0.97
0.94
0.95
197
55
142
2.3
2.6
2.1
0.9
1.6
0.0
0.97
0.92
0.96
1988-94
All EU-15 regions
Objective 1 regions
Other regions
1994-2001
All EU-15 regions
Objective 1 regions
Other regions
*= Beta is the parameter which measures the adjustment speed towards the equilibrium (or steady-state)
level of the per capita GDP of the EU regions.
Source: European Commission (2004), p. 146.
The question arises of how to establish investment priorities for supporting the goals of the
EU cohesion policy. Answers will differ according to the theoretical framework one is inclined to
subscribe. Under the endogenous growth theories, technology is the key factor23.
Economic structure is another obvious explanatory factor of productivity differentials, with a
clear disadvantage for regions specialising in low-tech sectors and agriculture (Bussoletti and
Esposti, 2004). Rural regions are scarcely populated and, sometimes, peripheral. The endowment of
traditional infrastructures, particularly transport infrastructure seems to be a convergence factor, but
results are not robust. New-economy infrastructures, such as telecommunication infrastructures
(TLC) and Information and Communication Technologies (ICT) are also correlated with regional
GDP growth regions (ESPON 2004).
According to the ERDF regulation, investment in environment and risk prevention should be
given a high priority in 2007-2013. The causal relationship between GDP growth and these
investments is however less strong than for innovation and infrastructure. According to some
literature, there is an inverted U-shape of the so-called Environmental Kuznets Curve. The
implication is that, for example, low level of pollution can indeed be found in both poor and rich
regions. In the former, because industry is less developed than agriculture, in the latter because of
country level, regional disparities and polarization (club convergence) have been increasing since the ‘80s (Quah,
1993).
23
According to Cappelen et al (1999, 2003), who regress GDP growth on a set of structural, technological and
“complementary-enabling” factors, across EU9 regions over the period 1980-94. R&D plays a particularly beneficial
role, but not so in the most under-developed regions. The same applies to the EU structural funds, which appear to work
better in the more developed regions (Cappelen, Castellacci, Fagerberg and Verspagen, 2003).
30
the high share of the tertiary sector and high-tech industries. Thus, regions at an intermediate
development level risk being particularly bad performers in terms of environmental protection24.
Summing-up, the empirical literature on regional growth in the EU does not point to strong
general regularities. Convergence seems to be conditional on a number of factors, which differ
among regional clubs. Traditional models in the Solow tradition do not work well, but the new
endogenous theories do not yet offer a consistent picture of the factors of growth. There is some
evidence, of course, that innovation activities and public infrastructures (traditional and new) are
beneficial to growth, but their returns seem to be very different across typse of investment and
regions.
It seems that growth models are still a weak foundation for investment planning.
Macroeconomic estimates can suggest the broad returns of infrastructure in terms of GDP or GDP
growth, but only as a part of a scenario analysis. A microeconomic approach seems to be a better
research avenue for establishing priorities in investment strategies. Cost-benefit analysis is, after all,
still the best tool we have, albeit an imperfect one25.
24
These predictions are however far from being proved in a convincing way (Panayotou 1993, Borghesi Vercelli 2003,
Shafik 1994). Recently the relations between growth, DDP levels and regional (mostly) variables in innovation,
accessibility and environment have been the object of a study ‘Policy guidelines for regions falling under the new
regional competitiveness and employment objective for the 2007 – 2013 period in the fields of the knowledge economy
and the environment, in line with the Lisbon and Gothenburg objectives’, prepared for the EC by an international team
lead by Csil Milano in association with EPRC European Policies Research Centre and FAI- Fraser of Allander Institute,
University of Strathclyde – Glasgow, DMIO, Department of Management and Industrial Organisation, Marches
Polytechnic University- Ancona and Mazars & Guèrard, Evaluation & Pilotage des politiques publiques-Paris.
25
Further research is needed to check the consistency of micro-data on infrastructure returns with estimates from
econometric data. One difficulty here is the lack of systematic data collection of the former.
31
3.
Public Investment Planning and Evaluation: Theory
3.1
Overview
Social CBA theory, after more than a century of evolution from its French origins (Dupuit
1844) and its first modern articulation in Cambridge, UK (Pigou 1947), is now based on solid
analytical foundations. These foundations are quite different from some popular views of CBA. The
most well developed theory now available is rooted in general equilibrium foundations, it includes
rationing disequilibria so common in the real world; it is welfaristic, but not necessarily
individualistic; it solves the planner’s problem at the same time in terms of shadow prices and of
feasible policies. All of this amounts to saying that most of the twentieth century experience with
CBA, based on partial equilibrium setting, market clearing assumptions, and narrow views of
government goals and constraints, is unnecessarily restrictive. There is, however, still much to do to
integrate this framework with the insights of incentive theory, the analytical framework that has
changed the perspective of public economics in the last twenty years.
This section discusses the relevant social CBA ideas in the context of a multi-government
setting. Drèze and Stern (1987, 1990) offer what is needed, a general theory (the DS framework)
that includes a number of earlier contributions as special cases. Laffont and Tirole (1993) and
subsequent work on incentive theory, including Laffont (2005), offer a frame for the analysis of
information and incentive structures (the LT framework). The two strands of literature in public
economics should be seen as complementary, something that is perhaps not yet fully acknowledged
in traditional CBA literature.
The section restates the DS framework in a multi-country setting and under incomplete and
asymmetric information. This is done in an informal way, albeit abstract 26. The shadow price
definition used here is the social opportunity cost of a change of the world. The economy I consider
includes different types of agents: individuals (consumers, workers, taxpayers), private and public
firms, regional governments, national governments, and one supra-national planner. They have
preferences and react to signals, such as prices and rations. Some of these signal are under the
control of the regional government, others under the control of other levels of government, other are
32
truly exogenous for all planners. Each planner computes its set of shadow prices, based on its own
preferences and constraints, and uses them to evaluate projects. The interplay of planners can be
modelled in different ways, including games, vector optimization, or a bottom-up lexicographic
approach. The latter means that when co-funding or licensing decisions involve different planners,
the projects selected are those mutually compatible, i.e. simultaneously passing a CBA test. This is
the ‘consensus decision set’. Under asymmetric information, however, one government or
implementing body may have an incentive to manipulate project evaluations to get co-financing
from the supra-national planner. Moreover, evaluation is costly. Hence, there should be a
mechanism to stimulate efforts of planners, evaluators and project managers, and to elicit true
information. This will be discussed in Section 4.
3.2
CBA theory in a multi-government setting
To evaluate projects in the public sector, including public-private partnerships, concessions or
subsidies to private firms, or other arrangements for the provision of public services, observed
prices are often misleading. Observed prices are almost always the result of the interaction of
imperfect markets and government interventions. Consequently, profitability at observed prices is
not sufficiently informative for planning and welfare analysis. We need more complete information
on the general economic impact of the project. This leads to the definition of shadow prices.
Shadow prices are the social opportunity costs of the inputs and outputs consumed or
produced by a project. They differ from observed prices because of market failures, e.g.
monopolies, public goods, externalities, or because of distortional policy tools, such as indirect or
factor taxes, rations, custom duties and quotas, public sector tariffs. This point is acknowledged by
several official CBA guidelines, including the CBA Guide, the UK Green Book, or the US Circular
A-94, but is often disregarded in practice for reasons to be discussed later.
In principle, any shadow price can be estimated as a weighted sum of the producer prices (net
of indirect taxes) and of consumption prices, with the weights given by the proportion of supply and
demand adjustments to meet the changes induced by the project, and in some cases corrections for
distributive elements. The logic of this proposition, frequently found in CBA literature, is the
following. If a project increases the demand for a production factor, e.g. electricity (and the supply
of an output) something has to change elsewhere in the economy to allow this. The change may be
given by decreased demand of the factor elsewhere in the economy because of the increased market
price of electricity, or by increased supply of it, again because of the price increase, or by a
26
Real word evaluation is necessarily distant from any abstract model. It is however, important to have in mind a
33
combination of these two adjustments. For small changes, competitive markets, and nondistortionary taxation, market (observed) prices equal the shadow prices.
For many goods, particularly capital, land, labour, services of utilities, non-tradable goods,
tradable goods under a quota regime, etc, this assumption is however often unrealistic. Thus we
need a theory of shadow prices27.
What follows is a non-technical, and highly selective, restatement of the DS framework28 and
its free adaptation to a multi-government setting. The economy I consider is described explicitly as
an array of governments and other social actors, plus a supra-national body. Thus, there are
different planners. Moreover, I stress some ideas that are particularly helpful in the context of
infrastructure planning in the EU and I show some implicit information and incentive issues in this
framework. The latter issues are important in a complex multi-government setting where there is
one player, the European Commission (or in fact a set of EU supra-national bodies), several
Member States, and lower levels of governments, the regions and other local actors. Probably this
setting has some relevance for other multi-lateral international aid organizations29, but I do not
develop here in detail institutional arrangements.
3.3
The economic environment
The abstract economy I consider includes a number of different agents. There are six types:
individuals, private and public firms, regional governments, states, and one supra-national planner. I
discuss here the first three types, and the latter, i.e. planners, in the following section.
a)
Individuals may be consumers, workers, tax-payers and share-holders. Workers are negative
consumers of their free endowment of leisure time. Some individuals may combine these four
aspects together, other may be just consumers (as children are), or consumers–workerstaxpayers with no shareholding, or consumers-taxpayer-shareholders, or other combinations.
Individuals earn different net incomes according to how much they work, their ability to
work, and the rents they earn through their shareholdings, the tax they pay or the public
transfer they receive (negative taxes). These individuals have well defined preferences, are
welfare (or utility) maximizers. Some of them may however be rationed, meaning that at the
general framework and then look for compromises with data limitations and other constraints.
27
In some cases, international or border prices, offer useful information for the estimation of shadow prices. This is the
most important message of Little and Mirrlees (1974), which however is a special case in the more general DS
framework, albeit an important one.
28
Drèze and Stern (1990a, 1990b) offer a less formal presentation of their theory. This is, however, perhaps still
difficult to grasp intuitively, as showed by the fact that several subsequent CBA textbooks still rely on traditional partial
equilibrium views.
34
observed prices they cannot consume what their income would allow them to do. Observed
prices are determined in different ways. Some of them are market prices, with some markets
competitive, other monopolistic or otherwise imperfect. The other prices are set directly by
planners, i.e. they are fixed tariffs. Prices are non negative by definition.
b)
Private firms are profit maximizer on behalf of their shareholders. Given their technologies,
prices of inputs and outputs, they determine their net supply, and earn profits gross of taxes.
Their net demand to the public sector is the difference between their total demand and their
demand to other private agents, i.e., other private firms, and workers. Firms fully owned or
partly owned by governments, but not under their direct control (i.e. not included in the
production plan of the public sector), are considered private. Thus, private agents are
individuals and private firms. Both have appropriate optimization programmes (respectively
for welfare and profit) constrained by resources constraints and any other relevant constraint,
and net demand or supply.
c)
Public firms, in contrast, are production units fully controlled by governments. Given their
technology, they need to efficiently produce what is requested of them by their principals, at
prices set by the governments (e.g. by a regulator). Profits, if positive, are cashed in by the
Treasury. Losses are covered by transfers. Privately-owned firms fully controlled by
governments are considered public ones. A generic name for a variable that influences private
agents’ behaviour is a signal. Signals may include, for example, producer prices, direct or
indirect tax rates, shareholding rights, rations on the production or consumption of specific
goods, transfers, and in fact any other variable that is needed to feed into the modelling of the
behaviour of private agents. Some of these variables are endogenous, meaning that their value
is determined by optimization. Others are exogenous, or parameters. All of these variables,
taken together, determine an economic environment. Private agents adjust their behaviour,
hence their net demands, to such economic environment.
3.4
Social planners, objectives and policy tools
Social planners have a specific social welfare function that embodies the objective of the
government. These include as arguments individual direct or indirect utility, but in principle could
also include other non-welfaristic variables, such as specific merit goods (for example expressing
political preferences or aversion to the consumption of some merit goods). Each government is a
welfare optimizer in this very broad meaning, and is budget constrained (perhaps not necessarily so
29
For example the World Bank uses a cut-off (social) discount rate in project evaluation for all countries, but in
35
strictly as to be unable to incur a deficit, but it certainly cannot select whatever deficit level it
wishes, as for example under the European Monetary Union rules).
This multi-government setting is particularly interesting because it is more realistic than the
earlier framework for planning in the Barone-Lange-Taylor tradition (Stiglitz, 1994), where there is
just one social planner. In that traditional framework the planner has all the relevant information
and powers, except the power to set ad personam optimal lump-sum taxation. In principle he can
either implement a command economy based on full control of volumes of inputs and outputs, or
decentralize the same Pareto-efficient equilibrium through a full set of shadow prices, and then let
the agents react to those signals.
When we have multiple governments, it may be the case that each of them has some powers,
but not all, in the very same economy. Regional governments, for example, may levy some type of
taxes (e.g. taxes on land) but not others (e.g. VAT). They may be in charge of the provision of some
services, e.g. health, but not of defence, etc. Thus the national or regional government may wish to
pursue a production plan or a marginal change of it (i.e. a public sector project) for the public
provision of certain services. To do so, it needs to disturb the economy, e.g. it needs to raise taxes
on land use, to hire workers for urban waste management. This relationship between the public
provision of services and the signals under the control of the planner is a policy function, or simply
a policy.
A feasible policy is such that, first, the scarcity or resource constraints are met, or in other
words that net private demands are equal to public production; second, the planner can select the
appropriate signals within its opportunity set. The latter means that the needed change in the signals
is legally or politically possible.
Regional or more local governments in some cases may have only one policy option: for
example the only way to increase the provision of some transport services may be through a tariff
such that the firm has a balanced budget (the Boiteux case). This is a very simple planning situation,
that will be discussed below. National government usually have a wider policy opportunity set than
regional ones, but not necessarily a wider production plan. Some EC Member States have widely
decentralized the provision of education, health, transport, environment and other functions to
regional governments, while have retained much larger tax powers. As a result, a system of
transfers is needed to finance the public provision of services at a regional level.
I introduce now the supra-national social planner. This planner, like the Member States and
regions, has its own social welfare function, a budget constraint and side constraints. Around the
principle one should compute and use both national and supra-national shadow prices, see next section.
36
world there are several supra-national organisations that have a quasi governmental status, but here
I am particularly interested in the European Commission30.
The EC has no tax power, and no production plan of its own (except for some public goods
such as legislation, information, and other intangibles). It does however have considerable
regulatory powers, and its own budget (based on transfers received from Member States). As a
policy tool the EC has some power to redistribute grants out of its own budget to Member States
and regions, targeted at specific interventions, such as infrastructure, human capital, assistance to
SMEs and R&D ventures, subsidies to agriculture, etc.
Consequently, the EC is of course a social planner, albeit with a relatively limited opportunity
set: it can disburse grants and regulate some production and consumption behaviour (e.g. it fixes
trade quotas with non EU states, establishes ceilings for some polluting activities, limits some
shareholding rights through competition legislation, etc.).
The acknowledgement that the EC is an independent social planner has some consequences
for planning and evaluation of public projects. This is the object of the next section.
3.5
Planning and evaluating
In the context of the previous section, planning has three different dimensions that may arise
in each government layer:
-
determination of the production plan,
-
policy selection,
-
calculation of shadow prices.
a) First, regions and member states need to establish their production plans, each for its own
competences. Production plan here means the quantity of the service to be provided by firms under
the control of government (publicly owned, private, or public-private partnership). Examples are the
amount of urban transport or refusal collection per year, or of water supply. At any point in time,
such production plans exist and can be observed.
The question arises whether we need to assume that public production plans are optimal, i.e.
whether the planner may not increase social welfare by a change in the vector of public output. The
answer is that, while the planner should try to optimise its choices, it is not necessary for the other
planning and evaluation activities to assume that the plan is optimal. The production plan may be
sub-optimal for several reasons, for example because the regional planner must comply with some
30
The EC is however only one part of the EU architecture, that includes the European Council, the various Councils of
Ministries, the European Parliament, and some EU judiciary institutions and consultative bodies, including one
representing the regions.
37
national or EU legislation; or for other political and managerial constraints; or for lack of
information.
Thus, we are interested in evaluating public production projects that represent marginal
changes in the existing production plan, even if it is sub-optimal. In fact, good projects are those
that change the production plan in the right direction, and no more than that is actually needed for
evaluation.
b) The next step is policy selection. This means that because we have a function that
associates a new production plan to a vector of signals under the control of the planner, there may
be more than one option open to the government. It is a great advantage of the DS framework that
policy optimization implies exactly the same evaluation process of the determination of shadow
prices. The intuition is simple. If the planner has to finance a production plan, he will select the
combination of taxes, rations, prices, etc. under its control that has the most favourable social
impact. If some of these variables are beyond his control, i.e. they are exogenous parameters, the
study of reforms can be pursued with a similar approach (an issue I am not going to discuss further
here, but that could be central in a serious regulatory impact analysis).
c) The third area of planning is the calculation of shadow prices. These have been defined as
the social opportunity cost of a marginal increase in the provision of one good in the public
production plan. Thus we have as many shadow prices as the dimension of the vector of goods
provided by the public sector, and their inputs (which are negative supplies).
In the next section I shall briefly dwell on shadow price rules 31. Here I wish to remain at a
very general level. By definition shadow prices are the impact on social welfare of a change in the
public provision of a good. If this change is small (in a technical sense), shadow prices are the first
derivative of the social welfare function around the optimum with respect to the good considered.
Equivalently, if the planner’s problem is expressed in the usual Lagrange or Kuhn-Tucker
constrained optimisation program, and constraints are expressed in the appropriate way, shadow
prices are equal to the multipliers of such constraints.
In an ideal world, the planners determines the optimal production plan, e.g. how much
motorway service is needed; then selects the optimal feasible policy to sustain such plan, e.g. taxes
on fuel; than it computes the shadow price of the motorway service, which may be different from
the observed toll of zero, or free access to the motorway.
It is at this stage, after planning is settled, that social cost-benefit analysis, or project
evaluation comes in. A new motorway project is proposed. A policy for financing motorways
31
I briefly discuss below the actual relevance of shadow prices for real world planners. Empirical estimates are beyond
the scope of this paper, but will be considered for future research.
38
should be decided. Then we need to evaluate whether that new motorway project is welfare
improving, compared with the do-nothing alternative, i.e. to stay with the existing stock of
motorways. To do so, we can either explicitly measure the net social welfare impact of the
increased service (the with-without project change of the world), or equivalently use the shadow
prices to compute shadow profits of the project.
We have one shadow price for the appropriate measure of the service (something like e.g. the
equivalent passenger per mile per year), and some shadow prices for the inputs, such as labour and
land. All the evaluator needs to do is to use these shadow prices in the accounting for the project
(including the consideration of any externalities) and use the simplest benefit-cost test: if the project
is profitable at shadow prices, it means that it increases the social welfare by exactly that amount,
and the project is approved. If the project has very limited shadow profits or incurs in shadow losses
(whatever its financial income forecasts), the project is rejected.
Having restated in this way a theory well established, but that is perhaps already not always
fully understood in applied CBA literature, what happens in the multi-government setting?
In the DS framework there is no all-powerful government. Each planner has limited powers,
but is consistent in planning and evaluation. Suppose then that, in one region, a motorway project is
proposed. Regional government has its own social welfare function and transport policy, and its
own specific constraints in the implementation of a transport plan. It will then evaluate the project
based on its own set of shadow prices, and this is the end of the story. Each government unit makes
its own project evaluation, based on its objectives, policy opportunity set, and constraints, including
those constraints arising from the decision of higher levels of government.
In the context of section 3.2, however, the evaluation problem may become nested in two
ways. First, one lower level of government may need the approval by another level of government,
for example national, and – second – may ask for supra-national co-financing.
There are then three planning and evaluating bodies that must cooperate in some way and in
principle because they face different objectives, policy mixes, and constraints, they will compute
different shadow prices for the same good32. Is this complex environment leading to intractable
problems?
This of course may be the case if, for example, two of the three players evaluate the project
positively, each based on its own shadow prices, and the third one rejects it because his evaluation
is negative. After all, income welfare weights, shadow prices of externalities, shadow wages may
32
An example is the social discount rate that does not need to be the same at EU level or for one specific Member State.
The same applies to several other shadow prices.
39
well differ when you move from a national to a regional perspective, and different decision-makers
have different views.
In general there are different approaches to solving this class of multi-agent problems, and I
shall discuss them in turn, albeit briefly and again informally.
a)
We can look at the problem as a lexicographic one. First comes the regional evaluators: they
select a number of motorways projects that pass the CBA test at regional shadow prices. Than
the national governments re-evaluate those projects, with their own national shadow prices,
and rejects some of them. Eventually, the EC evaluators re-consider the results, use their own
European shadow prices, and decide which one deserves a grant. This will be a subset of the
original regional selection, and one should only hope that this is not empty.
b)
Alternatively, one may think that we have three separate objective function, one for each
planner, and that we want to select those project that simultaneously optimize them. There are
two ways of doing that: either by a linearization of this multi-objective problem, which
implies giving weights to each objective function; or by other, more general programming
techniques (vector optimisation). In either cases a new shadow prices vector emerge and
should be used instead of any of the ones previously mentioned.
c)
Finally, one may think that each of the interested parties selects its preferred projects, and
then bargains (in some well defined way) to find out a commonly agreed solution. In this
game-theoretic framework the pay-offs to each player associated with project approval are the
welfare changes evaluated by their own shadow prices. The existence of one or multiple
equilibria is clearly highly dependent upon the specific features of the game, and there is no
general solution.
Which one of the three approaches is normatively the best avenue is difficult to say in
general, because there is here no definition of social welfare that is independent of the specific layer
of government and its shadow prices. Thus I will comment on the three alternatives on more
positive grounds.
First, the vector optimization solution is analytically attractive, but in fact it generates a fourth
(!) set of shadow prices, that may obviously be different from the regional, national and European
ones. Moreover, a kind of ‘constitutional’ agreement among the planners with (partly) conflicting
objectives is needed. In fact they should either agree on the weights to be given to their social
welfare functions; or they should agree to any other rule for vector optimization (there are different
ways of finding compromise solutions, and in such a selection process) there will be losers and
winners.
40
Second, the problem with the game-theoretic setting is that there is a very large number of
approaches to game design and solutions. No doubts some bargaining is observable between players
in real world situations, but from a normative perspective, it is very difficult to say which class of
games is more welfare improving in a well defined sense.
Thus, my preferred option, and in fact the one closer to the real world, is the simplest one, and
it is to think to the planning-evaluation process as nested and sequential, i.e. lexicographic. In fact,
if information on the shadow prices of the other two planners is known by one planner, the
sequence will be a purely virtual one. The regional planner will select its policies and compute
shadow prices, and will propose to national or supra-national planners only those projects that he
knows will pass their cost-benefit tests. After all, if the EC grants are additional finance, the
regional government, conditional to the national government’s approval, will fully finance the
projects that pass its own CBA test but not the European test, and will ask for co-funding for those
that pass both texts.
While the context I have described is purely theoretical, I think it has some practical
relevance, when shadow pricing rules are specified and made operational and information/issues
addressed. These are the object of the next sections.
3.6
Shadow pricing rules and their use
In the DS framework the roles of planners and evaluators are conceptually quite different.
Planners should carefully describe the social welfare function of the government, the signals
available as control variables, the signals outside the area of control, the scarcity and additional
constraints. Having set the objective function and all its constraints, they will calculate the first
order conditions of the constrained optimization problem. This involves calculating the derivative
of the Lagrange function to each control variable, i.e. the marginal social value of the control
variable, and deriving the shadow price rule by setting the marginal social value to zero. They will
then ask the evaluators to use these shadow prices in appraising specific public projects, to perform
the cost-benefit test on them, and then to select projects suitable for approval. In practice there will
be shortcuts, i.e. the calculation of proxies, but it is important to understand what the proxied
accounting price stands for.
In the multi-government setting I have described above, the EC, the national and regional
governments, each hire their own evaluators. The information on shadow prices relevant to each
planner is by assumption available to all evaluating teams. Thus, if the decision process is nested
through mechanisms of co-funding and/or of authorization, the projects actually implemented under
EU co-financing will only be those that pass, simultaneously or in sequence, three CBA tests.
41
Each evaluator needs here to be consistent in the use its own shadow prices. I shall come back
later on important information and incentive problems related to these multi-government setting. At
this stage, let us still suppose that information on estimated shadow prices and CBA evaluations for
individual projects are available to all players.
Having said this, the use of shadow prices by evaluators is quite straightforward. Evaluators
are ‘price-takers’, exactly like firms in a competitive environment. They calculate shadow profits or
losses and suggest a decision based on what basically is a forecasting and accounting job.
They will typically be interested in a relatively small set of shadow prices, for example:
۰
given a numeraire, the social discount rate, i.e. the shadow price of such unit of account
۰
the shadow wage rate, i.e. the social value of labour
۰
the shadow prices of other major inputs
۰
the shadow price of the output produced by the project
۰
the shadow price of some externalities and intangible goods33
۰
the shadow exchange rate and or a standard conversion factor for internationally traded goods
۰
welfare weights to account for distribution issues
Usually, they will not be interested in shadow prices related to other policy instruments,
because they focus here on evaluation of infrastructure projects. Evaluators of reforms, such as
reforms of trade, taxation, or environmental regulations will however focus on the shadow prices of
custom duties, of indirect tax rates, of pollution quotas, etc., but this is not our concern here34.
Planners have a significant workload in our environment, because in a sense they generate
new basic information. When calculating shadow prices and disseminating them, they act as
substitutes for the interaction among markets and government that generates observed prices, i.e.
the prices relevant to financial analysis in the private sector.
They should elicit from policy-makers, or interpret from available policy data (e.g. past
decisions), the social welfare function. This implies knowing important parameters, such as the
social welfare weights of individual welfare changes; the arguments of individual welfare and their
impact; the additional welfare impact of changes in any merit good in the objective function.
This is clearly a difficult task in practice, and CBA literature does not offer much guidance on
how to select the appropriate SWF. The DS framework assumes the conventional BergsonSamuelson SWF. This implies that shadow prices will assess the social impact of the project,
conditional to assumptions about the preferences of the policy maker on equity, as embodied in the
33
34
Typical examples are the value of time and the value of statistical life.
In the US most of the CBA debate in fact is about regulatory changes, see Adler and Posner (2001).
42
welfare weights. In fact, there are some useful reference in literature to distributional welfare
weights estimation (but far less practical experience).
A first consequence of the above setting is that in general shadow prices of goods will include
a distribution characteristic. This is different from other CBA approaches (see Harberger and
Jenkins, 2002), that propose using utilitarian functions (i.e. giving the same social weight to any
individual welfare change) and subsequently analysing the winners and losers at those ‘efficiency
prices’.
A second consequence of the DS framework is that general equilibrium effects enter in
shadow prices rules. The marginal social value of a good will depend on its direct impact on social
welfare and on the indirect impact on substitute or complementary goods.
Third, shadow prices and optimal policies depend on the planner’s interpretation of which
instruments are available, their constraints and exogenous variables. This point is certainly the most
difficult applied analysis task, because it requires detailed information on specific market
conditions. Just to give an example, the rules for the shadow wage rate (not just its value) will
depend inter alia on the specific constraints in that specific labour market. Thus, in fact, the planner
needs to have one empirical model for each market, often including the regional or local dimension.
Hence, the information needed to implement the DS framework is considerable, and it comes as no
surprise to find that nowhere has any attempt been made to use it in practice.
Three major departures from the DS framework can be observed in the real CBA world. First,
planners take shortcuts to estimate shadow prices, i.e. they resort to general, simple rules, which in
some case may be shown to be robust in different planning environment. This is the meaning of the
Little-Mirrlees rules, for example, and of several other project guidelines. Drèze and Stern (1987)
themselves discuss such shortcuts and their relative merit sympathetically. Second, the role of
planners and evaluators are often confused, i.e. project evaluators calculate shadow prices, because
planners do not provide them with their calculations, but with rather general rules. As a result,
inconsistent shadow pricing is more probable. Third, under multi-government planning and
evaluation, lower level evaluators often do not use their own set of shadow prices but (directly)
those of the higher level of government.
Moreover, as Little and Mirrlees (1994) discovered at the World Bank (and elsewhere, I
would suspect), only a limited set of projects are seriously evaluated by CBA, for reasons that are
only in part of technical nature. To understand why there is a strong disconnection between CBA
theory and practice, and to propose better evaluation designs, we need to look more in depth into
the information requirements and the incentive structure of the analytical framework.
43
4.
Information and incentives
4.1
Beyond the planning black box
I turn now to a more realistic view of the planning and evaluation mechanism, even if the
discussion is still quite general. In the standard DS framework the main information advantage as
compared with the Barone-Lange-Taylor model of planning is that there is no more an all-powerful
central planner, but a potentially large number of social planners. Each of them needs a fraction of
the information set, and can focus on the instruments available to him, and takes the other ones as
given.
While this is an interesting way to pick up the fragmented nature of many government
environments, where there are indeed many departments and agencies, regional authorities, and
supra-national decision-makers, the information burden is still considerable. First, governments
interact in the decision process. Second, the standard framework disregards possible effects of
uncertainty, information incompleteness and asimmetries, and incentive problems.
In what follows I introduce a preliminary discussion of some of these problems. The
discussion is going to be informal and generic, but it can be easily translated into a number of
game-theoretic models under specific hypotheses. The objective of this section is purely to illustrate
a research approach, more than to suggest solutions that can be immediately transplanted in the real
world. I hope, however, that future research in this direction can be helpful in the design of more
effective planning and evaluation mechanisms. Let us start with a discussion of some open
questions from the previous section.
a) First, consider planners35. They have to model the economy in a fairly detailed way,
including an appropriate identification of the relevant social welfare function. In practice, there will
be a lot of uncertainty surrounding the variables they observe. After all, they have to use empirical
data, drawing for example marginal costs data from input-output tables and demanding forecasts
from econometric estimates. Thus, the shadow prices they will compute will be statistics based on
sampling and estimation errors. In many cases, they will use shortcut estimates (i.e. they are not
44
solving a planning program analytically) and hope that their results are not too far from the true,
unknown shadow prices.
In the rest of the discussion I am going to use interchangeably the word ‘shadow prices’ for the
analytical solution of the planner’s optimization program, and for their operational estimates, more
appropriately defined as ‘accounting prices’. These, however, are clearly two different concepts.
It is of course possible to diminish the degree of uncertainty of the estimated shadow prices through
increasing the planner’s research costs and efforts. How much effort and time the planner should
spend in such activities clearly depends upon its reward for doing so. In the words of Little and
Mirrlees (1994), there are cost and benefits of cost-benefit analysis. Until now I have not taken a
strong position on the nature of the planner, particularly I have not said whether it is benevolent or
whether it has a private agenda.
I prefer here to consider it explicitly as benevolent, i.e. the planning unit is just an office of the
government, without its own private agenda, and the government itself has a social welfare function
that does not include the private objectives of the policy makers36.
Whatever the SWF, in principle the planner should spend resources in acquiring information
until the marginal benefits of this activity equal the marginal costs. Suppose, for example, that the
planner has estimated that the social discount rate is 5%. In fact, because of uncertainties
surrounding the variables involved in this calculation, it would be safer to say that there is a 90%
probability that the true SDR lies in a confidence interval, i.e. within the 4%-6% range.
CBA literature has not often considered accounting prices as stochastic variables, but of course they
are conditional to the expected values of the estimated parameters, even if, for practical purposes,
planners have to write unique figures in official project guidelines. Uncertainty surrounding shadow
prices may typically lead to under-spending in this area of analysis by governments. In practice
fixed budget for planning are allocated in the form of constraints on available personnel. Thus, in
general it is safe to assume that the planner has imperfect or costly information on the shadow
prices he disseminates.
b) Second, consider now the evaluators. Several different organisational arrangements are
possible. The planner may either ask an internal evaluation unit within its layer of government to
appraise the project; or it may ask the project proponent to hire an independent evaluator. In the
past the World Bank relied mostly on internal project analysts, now it frequently hires consultants.
35
In this paper there is no explicit discussion about the relationship between planners and policy makers. I understand
the planner as an office or a team in the public sector, that administers a given capital budget.
36
Others may dislike this assumption, and prefer different ones. This is, however, not a concern for us at this stage.
Later I will give an example of evaluation with private agenda and corruption.
45
The European Commission considers ex-ante project evaluations to be the responsibility of the
Member State, who in turn can delegate it to internal offices or again to consultants. In future, as
mentioned in section 2, EIB teams will be directly involved in project evaluation under the
Structural Funds on behalf jointly of Member States and the EC37.
In fact, it is not exceptional that the initial project appraisal is provided directly by the project
proponent, who has more information and also an obvious interest in complying with the
expectations of the decision-makers. I assume below that the project evaluator is a professional
team, with a private objective function, different from the one of the planner (SWF) and the
manager (profit). Again, project information search is costly. While accounting prices and other
macroeconomic information are freely available at this level by assumption, albeit uncertain, the
evaluator needs some micro and macroeconomic forecasting, and should use data that are available
at the firm and households level.
Moreover, there is here clearly asymmetric information between the evaluator and the private or
public firm that proposes the project38. Again the evaluator can spend money and personal effort in
extracting information, but only up to a point. If the evaluator is a private consulting team, or even a
public sector office where the time and effort of the personnel have a private opportunity cost, the
evaluator will stop the evaluation activities when it is deemed that one more hour of work will
decrease its utility. The evaluator can of course present to the decision-maker its cost-benefit results
in the form of stochastic returns, e.g. disclosing a probability distribution of the net present value.
However, the decision-maker cannot know how much effort there is behind these results, and how
credible they are.
As a consequence, the decision-maker faces double uncertainty: in fact, he is uncertain about the
credibility of the evaluation, because of both the residual uncertainty surrounding the shadow prices
and the asymmetric information about the quality of the evaluation. This will often undermine the
credibility of the CBA-based decision-making process, and will favour more traditional alternatives,
such as purely administrative and legal appraisals, and politically-oriented decisions-making, which
of course are not better, but are perceived as less costly and controversial.
c) Third, even with a benevolent decision maker, and a fully controlled planning unit, the
evaluators risk being captured by the project proponent. This may happen either because of outright
corruption or simply because the evaluator will find it convenient to trust the project proponent too
37
The EIB in last years has regularly advised the EC on major projects. In future there will be an even closer and more
structured collaboration.
38
In standard CBA theory public firms are under the full control of government, and respond to shadow prices rather
than to market signals. In practice, it is difficult to make this distinction.
46
much, and save effort. Again, because the decision-makers may suspect this is happening, they will
only partly trust the evaluators.
These points needs to be discussed in some detail, because while well known to practitioners, it has
never been raised in CBA literature, in contrast to its importance in regulatory economics. A full
consideration goes beyond this paper, and I start with a simple example.
4.2
The principal-agent framework: an example
The illustrative model I present here draws freely on Laffont (2005), see appendix 2, but it
departs from it for some key assumptions. The supra-national planner wishes to offer a grant to
promote regional development, a regional government wants to get the grant to finance an
infrastructure, operating under conditions of natural monopoly, e.g. a motorway or a water
purifying plant. To simplify the game I assume that: a) public funds, in the form of a grant, are
provided entirely by the supra-national planner (the first principal); b), a regional authority has
contracting/regulatory/evaluating tasks (a second principal), c) the management is the responsibility
of a beneficiary firm, either public or private (the agent). There are two principals, one agent, and
asymmetric information.
The following social actors are going to be considered:
1. The supra-national planner wants to maximize social welfare, cannot be corrupted (has no
private agenda, perhaps because he is not elected and responds to an inter-governmental system
of checks and balances), but he is uninformed of the true cost function of the project that
depends linearly on technology and managerial effort (see below). He can ask the regional
regulator to elicit ex-ante information or he can send an “ex-post” evaluator39. The evaluation
on behalf of the social supra-national planner is given by comparative statistics on the welfare
changes of the agents, through a social welfare function with suitable welfare weights.
2. The regional regulator/evaluator is an office or team appointed by the regional government, that
is responsible for passing the grant to the beneficiary, for selecting it by a contracting
mechanism, on behalf of the planner, and for performing ex-ante evaluation of the proposed
project; because the key cost data are private information belonging to the beneficiary firm, the
regional regulator is asked to perform an ex ante evaluation to check the project data; there are
two possibilities: the ex-ante evaluator can discover whether the grant beneficiary, i.e. the
In this paper I use “ex-post evaluation” in the meaning of an inspector who is sent after the contract is assigned to the
firm. In the evaluation jargon “ex-post” often means after the project completion, or the start-up of its operative life. For
infrastructure projects this is, however, a too long time span from the contract assignment. Thus, I prefer here not to
commit the planner to a precise timing, I just assume that there is a probability to send the evaluator at any convenient
time after the contract assignment.
39
47
manager, is highly efficient, or alternatively the evaluator is unable to discriminate between
highly efficient and less efficient types. The probabilities of the two outcomes depend upon the
efficiency and effort of the evaluator.
3. The managing body is a public or private firm under the responsibility of managers who want to
maximize their utility; the latter depends on the difference between the net grant (i.e. the grant
net of cost reimbursement) and the value of the disutility of effort needed to minimize the
infrastructure total cost; the infrastructure cost, in turn, depends on a technology and the
manager’s effort; while effort is the private information of the manager, his technology can be
discovered by either ex-ante or ex-post evaluation; total costs are always verifiable ex-post.
4. The consumer gets a surplus, and, to simplify matters, the infrastructure is not tolled, the output
produced is normalized to unity (hence the consumers’ surplus is evaluated by a known shadow
price for the service, as established by the supra-national planner)
5. The tax-payer finances the grant, plus any rent extracted by agents because of rent-seeking and
corruption, and because taxation is distortional, there is a shadow price for public funds.
A well known key-aspect of this game in the LT tradition, is that the infrastructure project cost
(the net present value of the investment and operating costs) depends linearly upon a technology
parameter (adverse selection) and a managerial effort (moral hazard). In other words, the lesser the
manager’s efforts to control costs, the greater the actual project costs. Suppose that the technology
variable can have only two values: ‘high efficiency’, and ‘low efficiency’ (in more general models
there are frequency distributions within a range of values), with a probability known to everybody.
Thus all the players know that there is say 0,30 probability that the beneficiary is of the high
efficiency type and 0,70 probability that it is of the low efficiency type.
While the technology parameter is exogenous to the manager (perhaps because of monopoly in
knowledge, or for other constraints), the resulting total infrastructure project cost is endogenous,
because by working hard the manager can minimize costs. This effort has a disutility value to the
manager (if the effort is measured in working hours, money disutility is a shadow wage of the
manager, often greater than the observed wage, particularly in the public sector).
It is worth noting that if the infrastructure project is not a revenue generating one, i.e. the
highway is not tolled, the manager needs to be given a full cost reimbursement, including normal
profitability. The same is true if the motorway is tolled, but tolls do not cover the total cost. In such
a case it is the net cost that needs to be paid out by the regulator on behalf of the planner. This total
cost can however be high or low because of the adverse selection and moral hazard parameters.
Thus the planner (and the regulator) faces a trade off. They can either pay the project cost and leave
48
no rents to the beneficiary, but in such a case the manager has no incentive to minimize costs, or
offer the manager a socially costly rent as an incentive to minimize costs. Leaving rents to the
manager is socially costly for two reasons: first because of the shadow price of public funds,
second, because given a reasonably egalitarian set of welfare weights, it is harmful for the median
taxpayer to transfer income to the manager.
The net welfare change for the representative consumer-taxpayer is the difference between the
consumer surplus of the service, and the social cost of the infrastructure, (at shadow prices)
including the rents, where these costs are augmented by the shadow price of public funds. The
corresponding welfare change for the manager is the difference between the incentive element in
the grant and the disutility of the effort. Remember that costs are fully reimbursed through the
grant. In other words, the grant is in two parts, one is just a transfer to the manager from the
taxpayer to cover costs, plus the incentive element, also paid by the taxpayer.
The change of the social welfare is the weighted sum of the welfare changes of the consumertaxpayer and the manager.
If the grant were to be decided by the supra-national planner case by case, and there were
complete information, then the regional regulator and the ex-post evaluator play no active role. The
regulator fixes the incentive exactly to the level needed to minimize costs and to extract the optimal
managerial effort, and the incentive will just match the marginal disutility of effort, so that the
manager has no rent (as said, normal profitability is included in the total cost). Under asymmetric
information in the LT framework it is well known that the planner is compelled to leave a rent to
the efficient type. In fact, the planner does not know the type of firm, and is constrained to offer a
contract that allows the participation of the less efficient type, i.e. a high cost contract. Given the
probability distribution of types, however, the manager granted the contract is, with a given
probability, more efficient, can mimic the less efficient type, and earn a rent.
Laffont (2005) goes beyond this standard framework, as he assumes that government has a
private agenda, and can extract a rent from the beneficiary. This additional rent extraction can be
costly, because it imposes extra-costs on the manager.
One example in my context is a regional government that pushes the manager to excess
employment when unemployment is high and artificial job creation is a popular policy. Moreover,
the regulatory office can be corrupted and accept some kind of hidden bribes from the manager (for
example the promise of future well paid post or other benefits).
Under this view, while the supra-national planner is benevolent by assumption and maximizes a
standard social welfare function, the regional regulator maximizes a welfare function that includes,
49
with an exogenous weight, the private benefits of the decision makers. Any private agenda welfare
weight greater than unity will be an indication of the degree of lack of democracy (i.e. of social
accountability) of the regional government.
In fact, the supra-national planner needs the regional regulators and the manager to implement
its development strategy, but is compelled to leave rents to them. How can the supra-national
planner limit these socially costly rents?
One mechanism is “ex-post” evaluation. With a given probability (e.g. for one out of three
projects) the planner can send an evaluator after the contract has been assigned. Each evaluation
project has a given fixed cost and allows for reporting to the planner the observed cost structure, i.e.
reveals with certainty the technology parameter.
A second mechanism is the contract between the planner and the regulator. The regulator,
working without additional costs as an office of the regional government, is responsible for ex-ante
evaluation. There is a given probability that the ex-ante evaluation will discover the more efficient
type technology, or that nothing will be discovered (the evaluation fails).
Consequently, the ex-ante information on the cost for the planner is the conditional
probability of the more efficient technology and of the probability of discovering it ex-ante. For
example, it will be just 0.25 if the probability of the two types and of successful evaluation are
respectively 0.50.
There is this sequence in the game:
1. The manager considers the project feasibility and discovers the technology he can use, high or
low, and this information is private
2. The regional regulator is responsible for ex-ante evaluation and either discovers that the
manager is highly efficient, or remains uncertain
3. The regulator, on behalf of the planner, offers the manager a menu of two contracts: one is a full
cost reimbursement for the lower technology, the other includes an incentive for a lower project
cost for the higher technology; the manager accepts one of the two contracts
4. If the manager accepts the cost-reimbursement contract, the planner can send (with a given
probability) the “ex-post” evaluator to check the technology parameter
5. If the manager accepts the low cost contract, then the “ex-post” evaluator is not sent, and the
contract is implemented
6. If the manager accepts the incentive contract, and the “ex-post” evaluator is not sent, the
contract is also implemented
7. If the “ex-post” evaluator is sent (some time after the contract is subscribed), and confirms the
less efficient technology, the contract is implemented
50
8. If the “ex-post” evaluation discovers an efficient technology, there is a penalty40.
The planner knows, consequently, that there is only a 0,25 conditional probability of
implementing an optimal grant, with no rents for the policy-makers and the manager. He has to
accept that with 0.75 probability the grant is “excessive”. Notice that under these simplistic
assumptions, a key question is the cost of “ex-post” evaluation. One may think that this cost is
relatively limited, and that a systematic “ex-post” evaluation activity can solve the problem. This is
however unrealistic, because there are indirect cost due to renegotiation, project delays, etc. Given
these and some additional formal assumptions, it is possible to design the optimal incentive
contract, and to compute the optimal frequency of “ex-post” evaluation. The latter is a function of
different parameters, including the shadow price of public funds, of the probability distribution of
the adverse selection parameter, of the disutility of effort for the inefficient type (for details see
Laffont, 2005, and Appendix 2).
Here the trade off for the planner is that either he allows high cost contracts or he needs a great
“ex-post” evaluation effort to contain rents. The no-incentive, low-powered contracts do not create
rents, neither do they minimize costs. Conversely, the incentive contracts generate rents, but
minimize costs, and do not need ex-post evaluation efforts.
Under some assumptions, the lower the inspection costs, or the extent of asymmetric
information, the higher the use of incentive contracts, and the optimal probability of ex post
evaluation. Conversely, the more inefficient and unreliable the regional regulator is, the higher the
shadow price of public funds, or the cost of the “ex-post” evaluator, the better is to offer a costreimbursement contract, hence fewer incentives and ex-post evaluation. In less developed regions
this case will be more frequent.
Notice that the existence of a private agenda of the regulator/ex-ante evaluator, implies a risk of
collusion with the beneficiary, to hide the signal about the adverse selection parameter. To buy the
loyalty of the ex-ante evaluator, the planner needs to pay a compensation proportional to the
information rent of the manager or – alternatively – there should be an strong and socially costly
enforcement mechanism. This additional cost enters in the social welfare function and acts as a
brake to the use of incentives. Inefficiency and the collusion of the regional government play the
same role, decreasing the use of otherwise useful incentive contracts, and pushing the planner to
offer low-powered cost-reimbursement schemes.
One conclusion of this simple illustrative example is that the optimal design of the planningevaluation-incentive mechanism for implementing the infrastructure plan is country or regionspecific, because it depends upon parameters that may vary greatly41.
40
A key-assumption here is that the threat of the penalty is credible. The penalty can be designed in different ways.
51
The key-parameters here are the social net project value at shadow prices, the shadow price of
public funds, the shadow wage of the management team, a reliability rating of the regional
regulator, an index representing the extent of asymmetric information. In a multi-government
setting, such as the EU, it is unlikely that the same planning design will be optimal for less
developed regions. In countries where there are unreliable regional regulators, high information
asymmetries, high potential corruption, high opportunity cost of public funds, it is probably difficult
to implement incentive infrastructure contracts. As a result investment costs will be high, rents will
be left to the potentially efficient types, etc.
In more developed regions, however, with a sound democratic environment that contains
corruption, where regional regulatory offices are efficient in their ex-ante evaluation activity, and
where there is widespread information on available technologies, incentive contracts can offer good
chances of decreasing infrastructure costs.
Having said this, in a more evolutionary perspective, the planner should invest in capacity
building and in democracy deepening in the less developed regions. In the Laffont (2005)
framework there is limited scope for progress. A learning mechanism is needed, however, in order
to promote something as a cooperative evolutionary game between planners, regional governments,
and managerial teams. Cost-benefit analysis can be seen as such language, evaluation contracts and
project performance mechanisms can be helpful in this perspective.
4.3
Linking evaluation and co-financing contracts
In the above setting the ex-ante evaluator has a role similar to a regulatory office for public
utilities. The important difference is that usually the regulators do not disburse grants and act
through other signals, such as permitted price increases in the price-cap system or in other schemes,
concessions of legal monopoly or otherwise, etc42. In the LT setting, however, the regulator pays
back to the utility the public service costs or a share thereof, and offers an incentive to cover the
disutility of effort. The typical regulatory problem in this context is that, while costs are verifiable
ex-post by the regulator (an “ex-post” evaluation in our setting), cost-reducing efforts are the
private information of the agent (moral hazard). With more than one agent and a distribution of
technologies, some more efficient than others, there is an asymmetric information problem. This
leads to opportunistic behaviour by the efficient-type, who can represent itself as less efficient.
41
This is in contrast with the current EU Structural Funds mechanism, where the grant formula is essentially the same
for all eligible investment types and countries.
42
In other words, you can subside an highway investment either by a capital grant or by allowing for tolls that include a
capital cost component, and a rent.
52
Thus the regulator faces a trade-off between incentives to raise efficiency, and rent appropriation by
efficient types.
The ex-ante evaluator has a similar problem. She or he is responsible for suggesting that the
supra-national or regional planning body offer a grant. The project proponent has a clear interest in
presenting the project as financially costly net of revenues, and economically beneficially net of
costs. Even if the proponent and the evaluator use the shadow prices given by the planner and the
market prices given by suitable forecasts, the true costs of fixed investment, personnel and other
inputs, including managerial efforts, are best known by the project proponent. Should the ex-ante
evaluator trust the information given by the latter, or invest its own resources in collecting
information? In general this would be very costly, and again some uncertainty will surround the
evaluation.
These issues are further exacerbated in a multi-government setting, where some strategic
manipulation of information – even among benevolent governments – cannot be ruled out. After all,
each government layer has different constituencies, and its social welfare function does not need to
include the welfare of citizens of other constituencies. The current experience around the world is
that these issues are very often not effectively addressed by governments and international
organizations. This is compounded with other sources of mistrust in economic analysis. The
calculation of shadow prices is costly, and planners tend to delegate it to project evaluators. This is
clearly a capital mistake in an incentive perspective, because if ex-ante evaluators work for a
regional government with a private agenda, or – even worse – for the grant beneficiary, they will be
under pressure to use shadow prices that lead to high economic returns43.
To sum up, while standard CBA theory virtually offers a solid framework for project planning
and evaluation, the information and incentive structure surrounding these activities is often such
that it undermines the credibility of CBA itself. Apart from any ethical and reputational aspects,
which are obviously of paramount importance for professionals in planning and evaluation, there
are also economic incentives that may play a role. These are at two levels: evaluators and firms (I
suppose, to simplify, that the social planner and the decision- makers coincide and are benevolent).
Evaluation can be seen as a contract between a principal, the decision-maker, and an agent,
the evaluation team. Moreover, public co-funding (ex-ante) project decisions should be based on a
combination of financial and economic evaluations, in order to capture the private and social impact
of projects, and may have an incentive element for (ex-post) high performing projects.
It is important to stress that the co-financing mechanism should be based on both dimension
of project appraisal: financial and economic. Private firms, public-private partnerships, or public
53
firms managed by profit-maximizing managers, are not going to sell or buy at shadow prices. They
are going to use observed or market prices. Thus their incentive structure is different from the
public sector one. For them, financial analysis of the return for the investor is all or nearly all the
project appraisal they need. Thus they are not really interested in economic analysis and in social
cost and benefits, if there is no specific incentive attached. Moreover, they are ready to manipulate
information provided to evaluators in order to obtain public sector co-funding, at regional, national
or supra-national level. This is not necessarily by cheating, it could just be a consequence of the
well known optimism bias of any investor, including public firms.
In most cases, however, firms are not going to implement the project without a public
subsidy. The most promising way to address the incentive issues I have described in this section is
to find a way to pay back to investors a financial reward for socially deserving projects, based on
ex-ante and ex-post evaluation, by eliciting true information through an appropriate revelation
mechanism. Thus, the co-financing contract should be designed to offer an incentive (or an implicit
punishment) according to social outcomes. The usual implicit long-term incentive for the evaluator,
within or even outside the public sector is a reputation bonus for evaluation tasks perceived as
successful. While this is sometimes enough to warrant good evaluation, in general this mechanism
is weak when organizations, not individuals or small teams, are involved. Reputation can be
indirectly related to future money rewards. One way to think in this direction is when there is a
selection process of evaluators that considers information on their track record of success and
failures. A first step, already experimented recently by the EC, DG Regional Policy, in other
context, is to perform a systematic quality assessment of evaluation reports44. A second step would
be to create and disseminate systematic information about the correlation between ex-ante and expost project evaluation. After all, if an ex-ante evaluator is discovered to be systematically overoptimistic, he is less credible than a more prudent one, and in an ideal selection/compensation
mechanism there should be a reward for evaluation efficiency and effort.
4.4
Using Project Return Benchmarks as Incentives
One practical approach worthy of further discussion to design incentive-based mechanisms is
to use return benchmarks. There are several ways to establish benchmarks by the planner, but the
most straightforward way is to systematically store information on ex-ante and ex-post project
returns. Based on long memory, the planner can calculate the median value of returns from the
43
Of course, they can also manipulate assumptions about timing and quantities, e.g. demand forecasts.
For example, a score can be given by assessors to the quality of various dimensions of the evaluation report, and an
overall rating can be computed.
44
54
relevant project portfolios. Projects that promise to outperform the benchmark should be given
approval, co-financing and a bonus, to be disbursed only if the expectations are confirmed after “expost” evaluation.
The implicit punishment for projects that do not achieve their targets is that their incentives
will be lost. Here the participation constraint is that the share of the contract in the form of incentive
should be not so high as to make the contract unattractive.
Let us see a simple worked example of this mechanism. The ex-ante evaluator appraises the
project and, based on available information, and after risk analysis concludes that the project
expects a (real) financial rate of return of 2% and an economic rate of return of 5%, using market
prices for the former and shadow prices, given by the planner, for the latter.
Suppose the benchmark for the financial return is 4%, and this is the financial discount rate,
hence, the project financial net present value is negative. A grant is needed to fill the funding gap.
Moreover, suppose that the benchmark for the economic rate of return is also 4%.
Thus, compared with the economic benchmark, the above project deserves the grant in social
cost-benefit terms. The planner then determines a co-financing grant in two parts: a funding gap
part, to cover the net present value of the project cost plus a linear incentive, proportional to the
ratio between the expected economic rate of return and the benchmark, which is 1.25 in our
example. Consequently, a fixed share of the excess social benefit translates into a financial rent to
the project proponent (or the regional government). Let us say that the fixed share is 0.5.
One way to design the incentive is to increase the financial discount rate correspondingly to
establish the project funding gap. In other words, the planner offers to pay the difference between
the net present value of revenues and costs, discounted at 4% financial discount rate, plus promises
to pay an additional bonus, equal to half the difference between using a 4% and a 5% in the
calculation of the grant due:
NPV (4%) + 0.5 [NPV (5%)-NPV(4%)]= 0.5NPV(4%)+0,5NPV(5%).
The grant is paid in two tranches. The first tranche is an ex-ante payment, based on the
normal discount rate. The incentive is set aside by the planner. Some years later (the timing of the
“ex-post” evaluation depends upon the project construction and start up forecast) the project is reevaluated45. The incentive is then disbursed to the managing authority responsible for the project
only if the expectations of ‘economic’ out-performance are confirmed, or correspondingly reduced
if they fall below the target.
45
Probably, for practical reasons, a fixed time span from the contract assignment is more advisable. In principle, the
optimal timing is however variable.
55
Let us see how the incentive works here. The first part of the grant is a cost reimbursement,
low-powered contract. With a given probability, it offers a hidden rent to the beneficiary, who has
an interest in exaggerating the project costs (or in hiding revenues). The second part of the grant,
gives probably an incentive to exaggerate social net benefits at shadow prices. The two biases,
however, work at least partially in opposite direction, because for a given set of shadow prices,
provided by the planner, exaggerating costs or under-reporting expected revenues, decreases the
incentive element. Moreover, because the incentive part is allocated ex-post, forecasting mistakes
have an opportunity cost for the beneficiary. On the same vein, one can think to incentives, beyond
reputation, for the “ex-ante” evaluation team46, when the project is discovered to be performing as
expected.
Again, this is no more than an illustrative example. Real world situation may need many
adaptations, particularly for administrative reasons. For example, project performance rating can be
based on more than economic returns, and translated into a qualitative scoring scale, to consider a
number of exogenous shocks in project implementation. The logic of the signalling mechanism
does not change if instead of one performance indicator (i.e. the economic rate of return) a set of
indicators is considered. Consistency is however needed. The project performance bonus, in
principle, should be shared among the regional regulator, the managerial team, and possibly the
evaluator (perhaps through a transparent reputation signal).
4.5
On the EU co-financing mechanisms
Building on the benchmarking approach, one can re-think the EC co-financing mechanism
under the Structural Funds in a similar way.
First, the planner should fix a benchmark financial rate of return ‘before’ the grants. It is not
surprising that many infrastructure projects in need of public capital are expected to have low
financial returns, sometimes a negative return (otherwise there would be quite limited justification
for public intervention). This is not, however, a justification for the EC to accept any ‘funding gap’
to be filled by the subsidy, up to the regulatory ceiling for the EU co-financing rate. The EU grant
should cover the NPV of net project costs in financial terms, arising from the fact that the required
46
For example, the ex-ante evaluation contract, can be offered by the regional regulator as a menu of alternatives to the
pre-selected evaluation team. The evaluation team can either pick up a contract where the price for the service is
proportional to the investment cost, that are supposed to be known in advance, with no performance incentive; or for a
fixed price plus an incentive for ex post-performance. If the ex ante evaluator concludes that the project is not to be cofinanced, he will be paid a fixed fee in any case. Obviously, if one thinks that evaluators always provide optimal effort,
there is no need to apply an incentive. This is an empirical question, more than just a matter of ethical or professional
standard.
56
rate of return for investors is positive. As said, by default, projects with negative financial returns,
e.g. typically environmental projects, or social infrastructures, should show high economic returns
(see below) to be accepted.
Second, for the calculation of the grant, in principle, one would need a standard financial
discount rate in real terms. This cannot be too far from the real marginal cost of the public-sector
borrowing-requirement. Under an EU common capital market, a benchmark minimum value could
be the return on long-term (e.g. thirty years) risk-free bonds, for example, the return of EIB bonds
in Euro. This is currently around 2%. Because of a still imperfect capital market, and looking at the
growth perspectives of the new Member States, a premium can be added on this benchmark. This
premium can be either assessed country by country, which would be better, but more complex, or
fixed by the calculation of an average premium over the specific benchmark for the Cohesion Fund
countries. It should not be considered, however, as a risk premium. Elsewhere (Florio, 2006) I have
proposed a 3.5% real financial discount rate for the EU. As compared with the current rules, the
proposal implies in fact a decrease in the average grant per revenue generating projects (and for a
given amount of the Cohesion Fund, the opportunity to invest in more projects).
Third, projects requesting for public grants should not be expected to have an excessive return
on investment, (let us say much greater than a real expected FRR/C=5%, i.e. after appropriate risk
analysis) otherwise one can assume that the private sector will be able to finance the project.
Moreover, one would expect that the return on national equity capital, FRR/K, or ‘after’ the grant,
to be higher than FRR/C, but not exceeding a given threshold, otherwise the project is asking an
excessive grant from the EU.
Fourth, and most important, development objectives (economic growth, environmental
sustainability, equal opportunities, distribution effects) are in principle measured by economic
returns, not by either FRR/C or FRR/K. Because externalities and shadow prices are explicitly
considered by cost-benefit analysis, in the ways suggested by the CBA Guide or in more elaborated
frameworks, most projects with low or even negative FRR/C may show positive ERR to be
deserving of a grant. Projects with a low ERR (based on the real GDP growth prospects in the new
Member States) should be considered unappealing by the European Commission.
Fifth, the EU grant should offer a reward to regional authorities, project managers and ex-ante
evaluators, proposing the most deserving projects in socio-economic terms. One simple way to do
57
so is to stipulate that if a project expects to outperform an economic benchmark, then the ratio of
the expected ERR and the benchmark is used as a conversion factor for the financial rate of return.
The incentive component should however be paid at a later stage, if confirmed by “ex-post”
evaluation.
Here the incentives to manipulate financial and economic returns, in general goes in the
opposite directions, because if the proponent, for example, exaggerates the demand for transport,
than he will obtain a lower financing gap against the incentive for the high economic return. Thus,
excess optimism in cost-benefit analysis would be partly self-correcting. However it may always be
the case that a project proponent tries to manipulate CBA, especially for the estimation of nonmonetary social benefits, such as, for example, positive environmental externalities. The latter,
while affect positively the economic rate of return, are irrelevant for the calculation of the EU
financing gap, in order to show high net social benefits of financially negative projects. This leads
to the final suggestion.
Sixth, lack of realism in cost-benefit analysis should be sanctioned by an ex-post penalty. In a
previous section I suggested offering this in the form of conditionality in the disbursement of the
reward. A part of the EU grant should be set aside in the first instance. The objective of this “project
performance reserve” would be to reward the financial analysis (where the proponent unavoidably
has some incentive to exaggerate costs) and economic analysis, (where there is the opposite
incentive) with a bonus for projects with a realistic financial analysis and high re-estimated returns
in terms of social costs and benefits. I suggest that some time after project assignment the EC
should appoint independent evaluators to rate projects according to a clear set of performance
criteria47. In other words the EU co-financing rate should have a fixed ex-ante component that
depends on the ‘funding gap’ under a standard financial discount rate (but with the above
mentioned limitations) and a variable component, that depends on the ‘relative’ economic rate of
return and can be disbursed as a bonus at a later date (e.g. after three years from the start up) if the
ex-ante social net benefits are going to be confirmed by this re-appraisal. Countries, regions or
sector managing authorities that systematically underperform according to their own economic
expectations will then be punished, because their bonuses will be reallocated to those who have
matched their cost-benefit targets48.
Evaluation contracts should be so designed as to offer a bonus for realism, as assessed “expost”, either in terms of reputation or perhaps of a direct monetary reward. Shadow prices values or
47
A rating agency would have useful side effects in terms of visibility and to ensure consistency. One regional
government and the public at large may thus see how many (“triple A”) public project it has been able to implement.
58
formulas should be agreed between the EC and other planning bodies, and not delegated to
evaluators. This will help all players to play the development game with common rules and to learn
from their success and failures.
48
Technical assistance should however help the less performing regions to investment failures.
59
5.
Concluding remarks
This paper is mainly methodological and unavoidably rather abstract. It should be seen as just
a starting point for new research focussing on cross-fertilisation between cost-benefit analysis and
incentive theory. It does not offer plug-in shadow prices, or cooking recipes for contractual
arrangements in the economics of infrastructure. Its objective is to offer a framework for discussion,
with particular regard to the future of the EU Cohesion Policy. There are four conclusions of the
above discussions.
1. Europe needs a huge investment effort for (broadly defined) infrastructure in the next decade, to
support the ambition of an integrated economic space from Bulgaria to Spain, or from Finland
to Malta. There is a limited world-wide experience of such a co-ordinated effort. In terms of the
financial resources involved and of the planning mechanisms, the experiences of infrastructure
funding by the US Federal Government or by the World Bank are more limited, and have a
quite different orientation. The most novel aspect of the EU experiment, which has slowly and
cumbersomely evolved over decades, is a complex multi-government planning mechanism,
involving the European Commission (and other EU institutions), national governments, and
regional authorities. There are two interwoven dimensions of this framework: new public
policies and financial mechanisms. The EU policies e.g. in transport, environment, energy,
competitiveness are pushing national and regional actors towards new challenging objectives.
Most of these objectives need investment to be sustained, and in turn investment needs finance.
A combination of EU grants, loans from the EIB and the EBRD, and their leverage effect on
private capital, is going to mobilise a huge amount of private savings, backed by the reassuring
environment given by the European Monetary Union. The core of this experiment is the
systematic exploitation of cross-boundary externalities, in terms of capacity building in the
public administration, of the learning processes of the many actors, of economies of scale and
scope in planning and financial packages. This public good dimension is very much
misunderstood by proponents of re-nationalization of the EU regional policy, who suggest that
the key issue is just to transfer funds to less developed Member States, and then let them plan
and manage their projects as they wish. My understanding of transferring national savings to the
EU budget is that this is not just a way to manage a multi-lateral development aid scheme, from
60
the rich to the poor member states. It is a way of creating an incentive mechanism in order to
build integrated economic plans, to serve the long-term EU policy goals. Under this view, the
design of a common European framework for investment planning and evaluation is the true
added value of the Cohesion Policy, of having a Trans-European Networks policy, of having
financial institutions as the EIB and the EBRD, and of several other EU mechanisms to support
investment.
2. One concern in this paper was how economists can contribute to this investment effort in a
meaningful way. Other social scientists may have a lot to say about, for example, the political
and legal framework needed. The failure of the European Constitutional Treaty is there to
remind us that the EU project is far from being achieved in terms of consensus, and we all are
aware of some nightmarish aspects of the current inter-governmental decision-making
mechanisms. Economists tend too often to oversell their recipes. They can, however, be helpful,
if they are clear about the boundaries of their knowledge. On the relationship between
infrastructure investment and growth, in a macroeconomic perspective, we must confess that in
spite of decades of academic research, the evidence is mixed. Modern growth theory, from the
Solow neo-classical model, to endogenous growth models, is an extremely active research field,
but its results are not robust enough to be used for actual investment planning. Growth models
can offer an insight into possible economic long term scenarios. Econometric analysis or inputoutput techniques can suggest some illustrative coefficient estimates. I would not, however,
suggest that a planner trust the high social returns of transport investments based on the
empirical elasticity in an aggregate production function as an indication of giving high priority
to motorways or railroads in the allocation of public funds. More research is needed about the
relationship between the macro and micro estimates of public investment returns. My tenet is
that, whatever the demand side effect of public investment, in the long term it is their intrinsic
social value that matters, and this is not captured by market signals in most cases. Well
designed, costly motorways that nobody uses are useless. While useless private investment is
wiped away by markets, this is not the case for useless infrastructure, which enter in the national
capital stock and stay there for decades, at their historical cost, and are slowly depreciated.
Conversely, still very useful public infrastructure, after conventional depreciation disappear
forever from national accounts. There is thus a fundamental methodological flaw in growth
accounting. By comparison, microeconomic social accounting, i.e. cost-benefit analysis, despite
its limitations, is more reliable as a support to investment planning. Macroeconomic models
should be used at a different level of economic policy-making, to build long-term scenarios, to
explore short run demand shocks, to evaluate fiscal policies.
61
3. Cost-benefit analysis is widely taught to economists at undergraduate and graduate level, and it
is has a long and distinguished intellectual history. It has occasionally attracted the interest of
the top of the profession, and is embodied in hundreds of papers and books. It is, however, still
frequently misunderstood in its theoretical foundations, and applied in an inconsistent way on
the ground. Both issues need to be addressed if CBA is to play a meaningful role in EU
investment planning. It is a great merit of the EC to have explicitly asked for CBA in the
Structural Funds regulations. The CBA Guide sponsored by DG Regional Policy was a step
forward, but it is aimed at non-specialists, and it offers a very simplified approach for desk
officers and consultants. In this paper I have tried to present in an informal way, i.e. without any
maths, the CBA theory in the DS framework, and to adapt it to a multi-government setting. The
key-message of this theory is that shadow prices are not proxies of market equilibria, but are
planning signals that solve a policy-constrained social planner problem. For example, the social
discount rate, does not mimic the equilibrium financial interest rate; the shadow wage is not the
supply-demand equilibrium price of labour; the value of time in transport or the shadow price of
carbon emissions are not given by virtual markets. There might be, or there must be shortcuts
and proxies, in practical estimation of shadow prices, but one has to know what the target of this
empirical estimation is. When the shadow price is defined in terms of the social opportunity
costs for the planner, one has to consider the dual dimension of public production and policies.
There is no such thing as a shadow price of healthcare if you do not define the supporting
policy, because the shadow price itself changes if public service provision is paid by tariffs, by
indirect taxation or in other ways. However you do not need to think in terms of an all-powerful
central planner to have a consistent theory of CBA, you can think in terms of multiple
governments, each of them using its own shadow prices in a consistent way.
Having said this, there are three concluding questions. First, do we really need shadow
prices in developed market economies, are markets not giving us sufficiently ‘right’ signals.
Second, if we need to use shadow prices, which amounts to saying that if we really need CBA, who
is going to compute and use them? Third, what happens in a multi-government setting, when
financing decisions by different social planners are linked each other? I have suggested the
following answers: a) yes, we do need shadow prices in the UK as we do in India, because market
failures are pervasive, policy interventions are widespread, and nobody can show that the value of
the statistical life or of passengers’ time (or of many agriculture and energy products) is proxied by
market signals in Europe, just because we have developed market economies; b) planners must
62
compute a set of shadow prices, evaluators should use them for project appraisal, and the two
functions should not be confused. In principle this distinction applies at each planning level, but a
consensus decision-set should emerge form this process, using a bottom-up approach; c) in a multigovernment setting there are information asymmetries that need to be addressed, and we should use
incentive theory, or CBA will be discredited: not because it is wrong, or too costly, but because of
the bad design of the planning and evaluation mechanism.
Incentive theory is at the forefront of microeconomic research for twenty years. Its core
message is that we have to look at the mechanisms that determine the actions or behaviour of
agents. For example, we cannot just assume that a firm will maximise profits or minimise costs. We
have to look into the black-box and discover that there are managers, share-holders and regulators,
each with its own agenda, and they interact. Incentive theory, while having had several founding
fathers, owes a lot to two French economists, Jean Tirole and the late Jean-Jacques Laffont, of the
University of Toulouse. They have deeply influenced industrial and regulatory economics, and I
suggest that, in the LT framework, infrastructure planning can be translated into a regulatory
problem. There is a European principal acting here as a planner, a second principal and a
national/regional government office, and an implementing agent, a private or public firm. The
planner pays a grant for infrastructure, can use ex-ante or ex-post evaluation, the firm knows its
own cost structure, and there may occasionally be cheating and even corruption. In this context
planning and evaluation are games that require good strategies and design to be played well. Here
CBA is the content of evaluation, it offers the right signals, but it is costly, it needs effort, and it can
be manipulated. I have tried to give examples on how to think parallel incentives for regional
planners, evaluators and project managers, assuming that the European social planner has no private
agenda (an assumption that I am ready to defend in relative terms).
My illustrative examples are relatively simple, even if unavoidably less simplistic than the
present funding mechanisms by the EC. Moreover, they may have to face opposition by those who
dislike incentive-based mechanisms, either because of political or bureaucratic constraints. There
may be a number of other formulas, to achieve higher efficiency in grants allocation, and several
political and administrative constraints that I have not discussed here. This was not the place,
however, to propose the details of financing mechanisms, but just a research perspective.
To sum-up: I propose moving away from the current low-powered incentive EU co-financing
mechanism, essentially an investment cost part-reimbursement scheme, towards a more incentivebased system, where there are the following desirable features:
63
1. infrastructure project proponents should be given an incentive to show that, while their project
has financial (net) costs they promise social (net) benefits: this incentive may take the form of a
financial bonus to pay for the effort of discovering socially beneficial projects
2. the effect of an economic performance bonus partly counteracts the perverse incentive of the
project proponent to show high investment costs (and low revenues), under any costreimbursement grant mechanism
3. The efforts of ex-ante evaluators should be enhanced by offering them reward for evaluation
effort and efficiency;
4. The planner should give the evaluators, regional authorities and the project manager the relevant
information about shadow prices to be used, and return benchmarks. Cumulative information on
projects should be used to establish benchmarks and learn about factors in expected
performance.
5. The regional authority should be involved in sharing the incentive mechanism with the project
manager
6. The incentive part of the contracts should be set aside by the planner, and confirmed, reduced or
cancelled, following ex-post evaluation
7. Ex-post evaluation is the responsibility of the planner, it should lead to re-estimation of
financial and economic returns at a fixed point of time after project approval, and to offer
performance rating. The information on the project performance rating for each regional
authority should be public. To ensure consistency in project rating, ex-post evaluation at EU
level should possibly be delegated to an agency with adequate professional expertise.
Having said this, a word of caution is needed. The application of the thresholds for returns
and the actual EU co-financing rate formula must not be too rigid. Under the proposed approach,
financial and economic rates of return should be considered jointly and in comparison with
sector/national or regional benchmarks, and ex-ante project analysis should be combined with expost evaluation. This mechanism is a way of playing the game, not an exact science. However,
games without clear rules are uninteresting. What matters, in practice, is the broad design and goal
of the mechanism.
I am confident that moving in this direction will be of some advantage for infrastructure
planning under the EU Structural Funds and the Cohesion Fund. The EIB and the EBRD, as other
financial actors, have their own objectives and traditions, but a closer dialogue among the EU
institutions on project evaluation methods seems useful, as in fact in several countries EU grant and
loan finance will be combined. The first steps have been already taken.
64
Finally, I suggest that there is a more general message arising from this EU story, something
that may be of interest in other contexts. Having sensible CBA rules printed in official evaluation
guidelines is only part of a more complex game between social planners and executing agents. Most
of the criticism of CBA among real world decision-makers arises from the less than careful design
of the interplay between different actors. What is needed is to move away from the naive view that
just because there are evaluation rules they shall be implemented. We need both dimensions, in a
nutshell: shadow prices and incentives. The design of planning and evaluation mechanisms is part
of a learning strategy by governments and development authorities, it should take advantage on
existing knowledge on project performance and failures, and should invest in offering the
appropriate rewards to the various agents involved in the planning and evaluation game. Other
international development actors should take notice of this aspect of the European experiment.
In a broader perspective, social CBA should be seen as a learning game among policymakers, planners, evaluators, project managers and the public at large, more than as an abstract
accounting technique. It is neither about market mimicking, nor a substitute for central planning.
Social accountability of cost and benefits, and the design of evaluation and decentralized planning
frameworks are parts of an open economic governance process. Democracy, in the old meaning of
public discussion, and not just of voting, is to be seen as the appropriate environment for the
scrutiny of social costs and benefits of investment projects.
65
66
Appendix 1.
Theory of CBA in the DS framewok
What follows is a short, simplified summary, with minor adaptations, of the general theory of
CBA in the DS framework, readers are referred to Dréze and Stern (1987,1990) for details. In this
framework CBA is a broad approach to social accounting for the evaluation of changes of the world
in terms of general equilibrium consequences. While public investment, particularly infrastructure
is the typical application, other fields of analysis include tax and regulatory reforms, for example of
environmental standards, rationing of private goods provided by the public sector, e.g. health and
education, subsidies and licensing of private investment, privatization, etc.
Definitions and notation
Planner: an office of government with a welfare function (SWF)
Project evaluation: the assessment of a project in social welfare terms
Shadow price: the net impact on SWF of a marginal increase of a variable (around the
optimum, a first order measure)
CBA test: accepts only projects which make profit at shadow prices
Policy: a rule that associates a state of the economy with a public production plan, Φ
Public sector: firms under full control of the planner (including optimally regulated private
firms)
Signals: s = (…sk…) k = 1,…K variables influencing private agents, e.g. prices, taxes,
rations
Environment: a vector s
Aggregate net demands E: s  E (s)
i : indexes for commodities (includes time, space, state of the world)
z: net supply of the public sector, z = (…zi…)
Scarcity constraint: E(s) – z = 0
Side constraint: s  S (S is the opportunity set of planner)
Project: dz, an infinitesimal perturbation of z (small project)
Feasibility: (z + dz)  Z
V: SWF including indirect utilities
Planning Problem E
All functions are once continuously differentiable.
V : s  V ( s) .
The objective function of the planner (SWF) is:
This is indirectly a function of behaviour of private agents, controlled by s
The problem of the planner is (P):
max V  V ( s)
s
E (s) – z = 0
sS
67
Policy
Φ(.) is a function that associates z  s
s.t.
(s, z) meets
sS
E(s) – z = 0
Example: for a public production plan, z there is at least one vector of taxes that are
compatible with public and private budget and other constraints. Once a policy is specified:
V  V [Φ ( z )]
sk  Φ (z )
To each production plan z we can associate an environment that, in turn, determines the state
of the economy, hence a level of social welfare. An optimal policy is such that (P) has a unique
*
solution s by assumption:
v
V ( z * ) V s

z
s z at the maximum of (P).
Project impact and evaluation
Given Φ, a policy, the social welfare change impact of a project is:
V Φ
dV 
dz
s z
V
V
 (...,
)
s
s k vector Rh
Φ k

 (...,
,...)
z
z i
matrix IxK.
The CBA test is: accept dz if dV > 0 (welfare improvement)
Definition of shadow prices vector
V Φ dV
v

s z
dz
Hence the CB test accept if П  vdz > 0.
П  shadow profits (given a feasible policy) and the welfare change is:
dV = vdz1 + vdz2 + + vdzi
In DS words:
“A project can be evaluated either valuing its inputs and outputs at shadow prices (dual method) or by
tracing all its general equilibrium effects, and then comparing the world with and without the project
from the point of view of social welfare (primal method)”.
Example: A new bridge: dzi, vdzi > 0, increases welfare when inputs costs (labour,
materials) and outputs (benefits) are evaluated at shadow prices v and net benefits are positive –
this is the Dual Approach.
The Primal Approach = V (z’) with bridge, V (z) without bridge, and V (z’) - V (z) = dv > 0
68
Decentralized planning
There are as many shadow prices vectors as independent planners (regions, ministries, states
in a federation). Each of them should select the optimal policy, given its SWF, and this determines
the optimal environment
s *  V ( s) *
z *  V ( z)*
Shadow prices depend upon the policy specified. By definition, the smallest area of control is
only one policy available. In such fully determined model there is only one feasible s for each z e.g.
*
only one set of public tariffs, indirect taxes and rations etc can support the production plan z .
Optimal production plan
The social planner should use CBA for arbitrary plans. Should he has sufficient information,
the optimal plan is the solution of the program (Q):
max V * ( z )
z
s.t.
zZ
(assume convexity of z)
v * z *  max v * z
zZ
V *
v 
z at
*
where
z *.
The following propositions holds:

Projects which display profits at shadow prices are welfare improving

Shadow prices are proportional to marginal rates of transformation in the public sector
and coincide for the optimally managed firm

Each public enterprise should maximize its shadow profits if there are no externalities
and this is a necessary condition for maximum aggregate public sector profits

The value (not the formula) of shadow prices around the optimum and those around an
initial arbitrary production plan are different.
Shadow price formulas: an example
To compute shadow prices objective functions and constraints must be specified.
Notation:
g = 1,
i = 1,
h = 1,
G
I
H
firms
goods
consumers
x  ( x , x ) : rations
x h (q, x h , m h ) : consumption plans
h
h

h

69
mh = money income
 r h   gh g
g
p = producer prices
q = consumer prices
t = indirect taxes, t = q - p
x   xh
xh  X h
h
and
rh = government transfers
θhg = → profit shares
1   gh   g
h
profit tax
The consumer problem
max Uh h ( x h )
q
x
h
s.t. qx  m h
xh  x h  xh
The producer problem
max  g  py g
y
s.t.


y g  Y g  y g : F g ( y)  0
yg  y g  yg
y   yg
g
: output
The Planner Problem specified
s : ( pi ), (r h ), ( xih ), ( yig ), ( gh )
Signals s include control variables (endogenous) and ω parameters (exogenous).
The private sector net excess demand is:
E ( s;  )   x h ( p  t , x h , m h )   y g ( p, y ) g
h
g
The planner objective function is:
V (s;  )  V (...V h ( x h ,...)....)
max V  V (.)
s
s.t.
x yz 0
sS
The F.O.C can be seen as the shadow pricing rule or the optimal policy rule. There are two
equivalent approaches to shadow pricing rules. Dual approach: indirect utility function and
uncompensated demand or Primal approach: expenditure functions and compensated demand.
Shadow prices and Lagrange multiplier (under some conditions, no side constraints)
70
V
E

0
s
s
Where λ is vector of a Lagrange multipliers and for a small project dz
E
ds  dz
s
The marginal increase of public production must be equal to the marginal increase of net
private demand
dV 
V
E
ds  
 dz
s
s
Lagrange multipliers coincide with shadow prices under some formal conditions and solve the
computational problem.
71
Appendix 2.
An evaluation model in the LT framework
The supra-national planner is a European SWF maximizer, who adopts an infrastructure
policy, offers shadow prices and grants to regional planners, appoints ex-post (contract assignment)
evaluators. The regional planner is a regional welfare maximizer, including the private agenda of
policy makers. He appoints ex-ante evaluators, selects projects, offers contracts to project managers,
The project manager is a profit maximizer, who wants to cover costs, earn a rent, knows
technology and knows how much effort is needed to decrease costs.
The ex- ante evaluator using regional shadow prices appraises a project producing one unit of good
with gross social benefit S and social cost C. S is fixed, C is variable. 
(adverse selection), e is managerial effort to minimize costs (moral hazard): C= e+. The latter
term is a vector of stochastic components , that is zero by assumption for the average project.
Technological costs are private information of the manager, there are two technologies, but
everybody knows their probability distribution: and (1. Managerial effort cannot be observed,
its disutility is e (fist and second derivative are positive). C can be always observed ex-post,
technology only when there is ex post evaluation.
The manager wants to cover costs, including normal profits, either by regulated tariffs or by
a cost-reimbursement mechanism, plus wants a rent. He asks an incentive to minimize costs, the
regional government offers a contract: C+ t.
where
There is no regional co-financing, to the EU taxpayer the social cost is (C+ t) (1+
 is the shadow price of EU grants.
The project is accepted if mutually compatible at EU and regional shadow prices. For the EU
tax-payer the welfare change is:
V = S - (C+ t) (1+
The utility change for the manager is: U = t - e).
Social welfare change, without welfare weights is: W = V+U , or
W= S - (1+e+t) + t -  e)
Equivalently:
W=S - (1+e+U +  e)) +U= S - (1+e+  e)) -U
Under complete information, the regional government acting on behalf of the supra-national
planner, extracts optimal effort, does not leave socially costly rents to the project manager, hence
will offer an incentive t= e*), such that U=0, and ’ e*)=1.
72
Under asymmetric information, the regional planner needs a truthful revelation mechanism.
One mechanism is here a menu of two contracts: C+ t, satisfying the incentive and participation
constraints, one for each values of 
The participation constraint for both management teams is U non negative.
The incentive constraint for the more efficient firm is that U of the efficient type is greater
than U of the less efficient type if he accepts the lower-cost higher –incentive contract than if he
accepts the higher cost-lower incentive contract. Symmetrically for the other type.
In this context it is well known that the efficient types gets an information rent in any case.
The rent can be limited by the ex-post evaluator, who is able to discover the technology parameter.
If the efficient type is discovered to have mimicked the inefficient type, there is a penalty. The
optimal frequency of ex-post evaluation is a function of its cost, and of other parameters, including
the degree of asymmetric information, the shadow price of public funds, the weight of the private
agenda of the regional policy-makers. For details, in the (somewhat different) context of regulation
and development, see Laffont (2005).
73
74
References
Abramovitz, M. (1986), “Catching up, forging ahead, and falling behind”, Journal of Economic
History, vol. 46
Adler M.D. and E. A. Posner, (Eds.) (2001) Cost-Benefit Analysis: legal, economic, and
philosophical perspective, Univ. of Chicago Press, Chicago.
Armstrong, H. W. (1996) “European Union Regional Policy: Sleepwalking to a Crisis”,
International Regional Science Review. Vol. 19 (3). p 193-209
Aschauer D.A. (1989),“Is public expenditure productive?” in Journal of Monetary Economics, 177200
Atkinson G., (2006) Environmental valuation and benefits transfer, Paper prepared for the V Milan
European Economy Workshop, University of Milan, May 26-27.
Barro R.J and X. Sala-i-Martin, (2004), Economic Growth, London, MIT Press
Barry, F. (ed) (1999) Understanding Ireland’s economic growth, New York: St. Martin's Press;
London: Macmillan Press. 242.
Barry, F., Bradley, J, and A. Hannan, (2001), “The Single Market, the Structural Funds and
Ireland's Recent Economic Growth”, Journal of Common Market Studies, Vol. 39 (3). P.537-53,
September.
Begg, I. (1999) “Previsiones sobre convergencia regional en la Union Europea” (with English
summary.), Papeles de Economia Espanola. Vol. 0 (80). p 100-122
Boldrin M. and F. Canova, (2003) “Regional Policies and EU Enlargement” CEPR Discussion
Papers 3744, CEPR
Borghesi, S. and A. Vercelli (2003), “Sustainable globalisation”, Ecological Economics, vol. 40.
Bradley (1997), “Evaluation of the Ratio of Unemployment Rates as an Indicator of Fair
Employment: A Critique” in Economic and Social Review Vol 28 No. 2
Bradley J, Morgenroth E.and Untiedt G. (2003) “Macro-Regional Evaluation of the Structural
Funds Using the HERMIN Modelling Framework”, ESRI Working Papers No. 152, Dublin,
September
Bristow, G., and N. Blewitt, (2001), “The Structural Funds and additionality in Wales: Devolution
and Multilevel Governance”, Environment & Planning Vol. 33(6). pp. 1083-99, June.
75
Bussoletti S. and R. Esposti, (2004) “Regional convergence, structural funds and the role of
agriculture in the EU. A panel-data approach”, Working paper n. 220, Department of Economics,
Marche Polytechnic University, Ancona;
Calderón C. and A. Chong, (2004) “Volume and Quality of Infrastructure and the Distribution of
Income: An Empirical Investigation” Review of Income and Wealth, Blackwell Publishing, vol.
50(1), pages 87-106
Calderón C. an L. Serven (2004), “The Effects of Infrastructure Development on Growth and
Income Distribution” World Bank Policy Research Working Paper 3400
Cappelen, A., J. Fagerberg and B. Verspagen (1999), “Lack of regional convergence”, in Fagerberg,
J., P. Guerrieri and B. Verspagen (eds), The Economic Challenge for Europe. Adapting to
Innovation Based Growth, London, Edward Elgar.
Cappelen, A., et al. (2003), “Regional support, growth and convergence in the EU”, Journal of
Common Market Studies, vol. 41.
Dauce, P., (1998), «L’evaluation des politiques communautaires de developpement regional:
Enjeux, methodes, resultats. L’exemple de l’objectif 5b en Bourgogne», Revue d'Economie
Regionale et Urbaine, Vol. 0(3) pp.379-94
De Rus G. (2006), Is Investment in High Speed Rail Socially Profitable? Paper prepared for the V
Milan European Economy Workshop, University of Milan, May 26-27.
Drèze J. and N. Stern (1987) “The theory of cost–benefit analysis”, in Auerbach A.. and M.
Feldstein. (Eds) Handbook of Public Economics, Vol. II, pp. 909–989. Elsevier North Holland,
Amsterdam.
Drèze J. and N. Stern. (1990) “Policy reform, shadow prices, and market price”, Journal of Public
Economics 42, 1–45.
ESPON (2004), Telecommunication Services and Networks: Territorial Trends and Basic Supply of
Infrastructure for Territorial Cohesion, Report downloaded from www.espon.lu.
European Commission, (2002), Guide to cost-benefit analysis of investment projects - Structural
Funds-ERDF, Cohesion Fund and ISPA, prepared by a group of experts for the Evaluation Unit,
DG Regional Policy, Bruxelles.
European Commission (2003), Annual Report of Cohesion Fund, Bruxelles
European Commission (2004), A new partnership for cohesion: Convergence, competitiveness and
cooperation. Third Report on economic and social cohesion, European Commission, Luxembourg
76
Evans, D. J., and H. Sezer (2004)” “Social Discount Rate for Member Countries of the European
Union” Journal of Economic Studies, 32 (1), 47-59.
Evans, D. J., (2006), Social Discount Rates For The European Union, Paper prepared for the V
Milan European Economy Workshop, University of Milan, May 26-27.
Fayolle, J., and J. Le Cacheux, (1998) «Elargissement, PAC, politiques structurelles et "juste
retour": la quadrature du cercle budgetaire europeen», Revue de l'Ofce, Vol. 0(66), p. 37-60, July
Fayolle, J. and A. Lecuyer, (2000), “Croissance regionale, appartenance nationale et fonds
structurels europeens: Un bilan d’etape”, Revue de L’Ofce, Vol. 0(73), p. 165-96, April
Florio, M. (1999) “An international comparison of financial and economic rates of return of
development projects”, University of Milan - dept. of Economics Working Paper, n.6
Florio, M. (2000) “Review of: Sturm J-E.: Public capital expenditure in OECD Countries: The
Causes and Impact of the Decline in Public Capital Spending”, The Economic Journal, 110, 464
June
Florio, M. (2001) “On cross- country comprarability of government statistics: public expenditures
trends in OECD national accounts”, International Review of Applied Economics, Vol. 15, N°2.
Florio, M. (2005) “A ‘European consensus’ on the growth agenda? A discussion of the Sapir
report”, Transition Studies Review, 12, N. 3, December pp. 411-4
Florio, M. (2006) “Cost benefit analysis and the EU cohesion fund: learning from experience”
Regional Studies, 40, 2, April
Florio M., Vignetti, S. (2003) “Cost-Benefit Analysis of infrastructure projects in an enlarged
European Union: an incentive oriented approach”, University of Milan - dept. of Economics
Working Paper, n.13
Florio M., Vignetti, S. (2004) “Cost-Benefit Analysis, development planning and the EU Cohesion
Fund: learning from experience” University of Milan - dept. of Economics Working Paper, n.31
Florio M., Vignetti, S. (2005) “Project analysis and European regional development: the EU
Cohesion Fund”, The ICFAI Journal of Environmental Economics, Vol III, n. 3, August.
Harberger, A C. and G. P. Jenkins (eds.) (2002): Cost-Benefit Analysis, Cheltenham/Northampton,
MA: Edward Elgar Publishing
Johansson P.O. (2006), On the use of the Value of a Statistical Life in CBA, Paper prepared for the
V Milan European Economy Workshop, University of Milan, 26-27 May.
77
Kubo, Y. (1995), “Scale economies, regional externalities and the possibility of uneven regional
development”, Journal of Regional Science, vol. 35
Kula E. (2002) “Regional Welfare Weights in Investment Appraisal – the Case of India”, Journal of
Regional Analysis and Policy, 32 (1), 99-114
Kula E. (2006a) Social Discount Rate In Cost-Benefit Analysis -The British Experience And
Lessons To Be Learned, Paper prepared for the V Milan European Economy Workshop, University
of Milan, 26-27 May.
Kula E. (2006b) Income Distribution In Cost Benefit Analysis By Using Welfare Weights, Paper
prepared for the V Milan European Economy Workshop, University of Milan, 26-27 May.
Laffont J.J. and J. Tirole, (1993), A Theory of Incentives in Procurement and Regulation,
Cambridge (Mass.), MIT Press
Laffont J.J. (2005) Regulation and Development, Cambridge University Press
Little I., and J.A. Mirrlees, (1974) Project Appraisal and Planning for Developing Countries.
Heinemann, London.
Little I., and J.A. Mirrlees, (1994) “The costs and benefits of analysis: project appraisal and
planning twenty years on”. in: Layard, R. and S. Glaister (Eds.), Cost–Benefit Analysis. Cambridge
University Press
Lolos, S.E, and A. T. Theodoulides, (2001), “The role of EU Structural Funds in the Development
of Lagging behind European Regions”, Economic & Financial Modelling, Vol. 8(1), pp. 29-46,
Spring.
Lolos, S. E. and Zonzilos, N. G.,(1994), “The Impact of European Structural Funds on the
Development of the Greek Economy: A comparison of two Models”, Economic & Financial
Modelling, Vol. 1(2), pp. 87-104, Summer.
Mairate A. (2006), “Does cohesion help economic growth?”, University of Milan-Dept. of
Economics Working Papers, n.6.
Nicolaides, P., (1999), “The Economics of Enlarging the European Union: Policy Reforms v
Transfers”, The Cyprus Review, Vol. 11 (1). pp.97-108. Spring.
Panayotou, T. (1993), “Empirical tests and policy analysis of environmental degradation at different
stages of economic development” World Employment Programme Research, Working Paper n.
238, ILO, Geneva
78
Payne, D., Mokken, R., and F. Stokman, (1997), “European Union Power and Regional
Involvement: A Case Study of the Political Implications of the Reform of the Structural Funds for
Ireland”, Aussenwirtschaft, Vol. 52(1-2), p. 119-49, June.
Pereira, A. (1997) “Development Policies in the EU: An International Comparison”, Review of
Development Economics, Vol. 1 (2) pp 219-35 June
Pigou A., (19473), A Study in Public Finance, MacMillan, London
Quah, D. T. (1993), “Empirical cross-section dynamics in economic growth”, European Economic
Review, vol. 37
Quinet E. (2006), Cost Benefit Analysis in France: Progress and Shortcomings, Paper prepared for
the V Milan European Economy Workshop, University of Milan, 26-27 May.
Romer, P. (1986), “Increasing returns and long-run growth”, Journal of Political Economy, vol. 94
Sapir A. et al, (2004) An Agenda for a Growing Europe - The ‘Sapir Report’, Oxford University
Press
Scapecchi P. (2006), Shadow Prices, the Environment and the Value of “Statistical Life”, Paper
prepared for the V Milan European Economy Workshop, University of Milan, 26-27 May.
Sezer H. (2006), The construction of social welfare weights based on equally distributed equivalent
income, Paper prepared for the V Milan European Economy Workshop, University of Milan, 26-27
May.
Shafik, N. (1994), “Economic development and environmental quality. An econometric analysis”,
Oxford Economic Papers, vol. 46
Spackman M. (2004), “Time discounting and of the cost of capital in government” Fiscal Studies,
vol. 25, issue 4, pp 467-518
Spackman M. (2006), Social Discount Rates For The European Union, Paper prepared for the V
Milan European Economy Workshop, University of Milan, 26-27 May.
Stiglitz J.E., (1994) Whither Socialism? Cambridge, Mass.: The MIT Press
Sturm J-E., (1998) Public Capital Expenditure in OECD Countries. The Causes and Impact of the
Decline in Public Capital Spending, Edward Elgar, Cheltenham, UK and Northampton, MA (USA)
Swinnen, J.F, (2001) “Will Enlargement Cause a Flood of Eastern Food Imports, Bankrupt the EU
Budget, and Create WTO Conflicts?” Eurochoices Vol. 0(0), pp.48-53 Spring
Welfens, P.J. (1999) EU eastern enlargement and the Russian transformation crisis, Heidelberg
and New York: Springer
79