The Productivity of the Public Sector: A Classical View

Marcella Corsi and Carlo D’Ippoliti∗

Abstract
We define public administrations (PAs) as productive units and analyze the recent trends in PAs’
productivity. We apply Sylos Labini’s productivity function, i.e. a Classical-type model of the
determinants of growth at both micro and macro levels. Such a framework is useful as a reference
growth model that does not rely upon “marginalist” hypotheses (especially concerning the aggregate
production function), and it is specifically well suited to capture the relevance of innovation processes.
We present a preliminary application of the model to the estimation of the impact on productivity of
the development of e-Government processes in a number of OECD countries. At the microeconomic
level, we propose a model of government action and of the production of public goods. At the
macroeconomic level, we set out to underline the relevance of distinguishing the public and the private
sectors in growth accounting exercises, and to highlight a significant relationship between e-
Government and economic growth.

Keywords: public administrations, productivity growth, e-Government

JEL-codes: H11, H83, L32, E12

1. Introduction
Public intervention in the economy is usually seen either with suspect, by libertarians and
neo-liberals, or as a way to fill in market’s gaps, by the relatively more progressive thinkers.
In the face of the current financial and economic crisis, in several (mostly European) countries
austerity measures are dramatically cutting on public expenditure, as if it was a deadweight
burden upon society, or a luxury that we can afford only in good times.1 Indeed, in particular
the role of the public sector as a producer of public goods and services is under attack, as
evidenced by the large role of public sector wage and employment reductions as a share of the
total expenditure cuts within the current fiscal retrenchment plans in Europe and elsewhere.2
In this paper we claim that the production of public goods is instead a constituent component
of any well-functioning capitalist economy and that it carries economic worth despite this can
only inadequately be measured by the market value of its output.
We do so by defining public administrations (PAs) as productive units, and by analyzing the
recent trends in PAs’ productivity. To carry on such an analysis, we apply Sylos Labini’s
productivity function, i.e. a Classical-type model of the determinants of growth at both micro

∗
Sapienza University of Rome, Department of Statistics, Viale Regina Elena, 295/E, 00161 Rome, Italy. E-mail:
marcella.corsi@uniroma1.it, carlo.dippoliti@uniroma1.it.
1
Cf. Kregel (2011); D’Ippoliti and Roncaglia (2011).
2
See Glassner (2010) for a review of the recent employment and wage reductions in European countries.
and macro levels.3 Such a framework is useful as a reference growth model that does not rely
upon “marginalist” hypotheses (especially concerning the notion of aggregate capital and the
neoclassical production function), and it is specifically well suited to capture the relevance of
innovation processes.
In the applied part of the paper, we present a preliminary application of the model to the
estimation of the impact on productivity of the development of e-Government processes in a
number of OECD countries.4 We employ a unique dataset developed for the European
Commission, estimating the value and structure of public expenditure for ICT-led
reorganization of PAs.5
The aim of the paper is thus twofold: at the microeconomic level, we propose a model of
government action and of the production of public goods; at the macroeconomic level, we set
out to underline the relevance of distinguishing the public and the private sectors in growth
accounting exercises, and to highlight a significant relationship between e-Government and
economic growth.
Such an analysis, however, incurs in the major limitation that the mainstream definition of
economic growth currently only takes the variation in GDP as an indicator, thus limiting itself
to the changes in sum total of the exchange value of the goods and services exchanged in the
market. Such a narrow view of economic worth has been repeatedly criticized.6 What is
fundamental from the perspective of valuing the public sector’s output is that publicly
produced goods and services are not exchanged in the market, they frequently cannot be given
an exchange value, and when they can such value if often an underestimation of the real worth
of such output for society. Thus, in the next section of the paper we consider what limitations
are inevitably occurred into, given the current state of the statistical practice, when applying
the concepts of efficiency and efficacy to the public sector.
The following section 3 illustrates the theoretical model we propose, i.e. an adaptation of
Sylos Labini’s “productivity function” model. In section 4, we highlight the potential of ICT-
enabled reorganisation of the public sector as a driver of productivity growth. In section 5 we
estimate the theoretical model and the specific role of e-Government processes in a number of
OECD countries; section 6 concludes by highlighting a few implications of our analysis for
public policy.

2. Limitation of the estimates of public sector output

Intervention of the public sector in the economy does not take the form of a one-dimensional
phenomenon: as it is attributable to a multidimensional world, given the multiplicity of
objectives it pursues and the instruments it applies, it clearly merits a multidisciplinary
approach, plural in both methods and objects of examination. Thus, economic analysis of the
productivity of the public sector, as sought in the present paper, inevitably provides us with a
partial picture. Moreover, the objectives of public intervention - of an essentially political

3
See Corsi and Guarini (2007) for a full description of Sylos Labini’s productivity function.
4
In accordance with the economic and organisational literature, by e-Government we mean a set of processes of
reorganisation and modernisation of the public administrations led by the adoption of Information and
Communication Technologies (ICT).
5
See Corsi et al. (2006). Carried out under the European Commission’s “Modinis Programme,” and managed by
the e-Government unit of the DG Information Society and Media, the e-Government Economics Project (eGEP)
focused on producing a measurement framework for the evaluation of e-Government impacts and outcomes. Full
documentation is available under request.
6
Lastly by the so-called Stiglitz-Sen-Fitoussi Commission’s Report: cf. Stiglitz et al. (2011).

2
nature - are neither constant over time nor necessarily uniform between countries, which sets
limits to the interpretation of international comparisons as in the case of the present work.
However, the most serious limitations to any quantitative evaluation of the economic role of
the public sector lie in the statistical practice related to the measurement of the public sector’s
inputs and output. Yet, it is an aim of the present work to show that at least some of these
limitations can be overcome, while others do indeed imply a high risk of underestimating the
real value of the public sector’s output but still they cannot impede us from analysing it
altogether.
For example, in the specific case of e-Government, it is to be pointed out that a series of
further effects stemming from its implementation are not taken into account in the estimates
presented below, although they have significant impact on the entire social fabric. Among
them, we may distinguish: increases in the level of responsibility and transparency within the
public administrations, improvements in the diffusion and circulation of information deriving
from public sources, greater participation in the performance of democratic processes,
enhanced efficacy for public policies. Yet, if we were to abstain from any quantitative
analysis only because it is to be expected that its results would be underestimated, we would
seriously incur in the more dangerous risk of implying that what cannot be measured does not
count.
For this reason, in this section we expose the major limitations of any quantitative analysis of
the public sector, in order to be then able to proceed keeping such limits in mind.
Most existing analyses of public sector productivity are defined at the microeconomic level,
in relation to an individual organisational unit within the PA. Occasionally, analyses of the
outcomes of the public sector production are defined or estimated at the aggregate level, as
are most of the feasible measures of efficacy (based, for example, on social-economic
development indicators as “social inclusion” or “health” levels)7. Although the aggregate
measures derive ex-post from the sum of microeconomic variables, this distinction takes on a
certain significance since ex-ante the macro magnitudes can differ considerably from a simple
sum. In fact, all the economic activities are connected, and a change within one organisational
unit cannot occur without producing effects within other, associated units (e.g., if a public
administration shows increased levels of efficiency, the benefits deriving from it are very
likely to be absorbed to some extent by the administrations interacting with it).
Furthermore, the operation to aggregate diverse magnitudes – in our case a miscellany of
goods and services – implies the need to adopt a common unit of measurement. For the
private sector the national accounting standards take market prices as a reference, aggregating
their value. However, the public sector has no market to sell its products and services, which
makes measurement problematic, and above all many public goods and services have no
market value: the conceptual implications are indeed considerable.
With regard to the former type of problem, it is to be noted that many public administrations
do not engage in the supply of services to final users, interacting solely with other
administrations (government-to-government activities): thus their place in the capitalist
economy is only on the side of input acquisition; many charge no price for the services
supplied, or charge only minimal amounts, in order to ration demand (thereby selecting
among a great number of consumers those who really need the services in question) rather
than covering, even partially, the costs borne. As concerns the second type of problem, it
should be noted that a number of public administrations have the precise aim not to supply a
certain service (for example, the PAs operating at the level of prevention – of certain

7
For a recent review see Giordano and Tommasino (2011).

3
behaviours and actions on the part of the citizens, for instance, or natural events, or even
threats from without). Furthermore, we are all well acquainted with the issue of “public
goods” – goods distinguished by collective and/or non-rival consumption, consumption of
which cannot be excluded for any individual user (as in the case of infrastructures): such
goods are generally provided without any charge.
No less significant, finally, is the lack of a clear and commonly accepted definition of the
public sector output, and of a value attributable to it. It is, indeed, precisely the many
conceptual difficulties and problems of definition that make measurement such a formidable
task. The solution most often adopted involves classifying as “market” activities, and so
aggregated on the basis of the payment (“price”) made for the individual transactions, that
part of the supply of goods and services to final users, which is acquired at a price amounting
to at least 50% of the unit production cost. Conversely, they are considered “non-market”, and
thus valued at the cost of production, all the remaining activities, namely those that do not
imply individual transactions, imply transactions only between PAs, or for which the charge
effectively paid is less than 50% of the average cost.
Such practice proves quite in contrast with the main aim of the empirical analysis here to be
performed, in that imposing a condition of equality between costs borne and value of output
produced is tantamount to implicitly assuming constant average productivity.8
The so-called Atkinson Report (NSO, 2005) has impressed on national statistical institutes
worldwide the need to address the definition of common public sector output measurement
criteria with all due commitment. The aim is to estimate this magnitude through direct output
measures, or in other words direct measurements of variations in the volume of output,
proxied by variations in the volume of activities and tasks pursued. With regard to progress
towards this goal, the Atkinson Report points up the pioneering position of the United
Kingdom within the EU, with estimation of about two thirds of the total economic activities in
the public sector applying direct output measures, while many OECD countries are lagging in
the implementation of this methodology with all the consequent problems when it comes to
comparing results.
In the last decades, several economists have attempted to assess the performance of the public
sector trough productivity indices that compare aggregate output to aggregate input use
(O’Mahony and Stevens, 2003; Dawson et al., 2005; Stevens et al., 2006). The use of direct
output measures is naturally easier in some industries, among which the health-care sector.
Beyond the cited Dawson et al. (2005), reviews of the economic studies focussing on this
sector are to be found in Cutler and McClellan (2001) and Cutler and Berndt (2001). Finally,
concerning the USA, particular attention to measurement issues in examining trends on ICT
usage and their effect on public sector productivity is given by Lehr and Lichtenberg (1996)
and Lichenberg (1996). In these papers the authors use data from the Bureau of Labour
Statistics’ (BLS) Federal Productivity Measurement Program on productivity growth and
computer assets. However, due to a lack of relevant aggregate data, our study will not
consider the USA.

3. A model of productivity growth

The economic model presented here constitutes extension and adaptation to the public sector
of the productivity function model by Paolo Sylos Labini (1984, 1985) for description of

8
For example, a pay rise in certain PAs of the public sector is evaluated sic et simpliciter as a proportional
increase in the respective production.

4
economic growth in the private sector (revised and estimated on Italian data by Sylos Labini,
2004), which is also the source for denomination of the individual effects.9
Let YPS denote the value of production in the public sector – the sum of the goods and
services supplied – calculated in monetary terms, which by definition represents a portion of
the Gross Domestic Product (YPS = GDP). Thus the productivity of labour in the public
sector can be defined as the overall value of production in the public sector divided by the
number of employees in that sector.

Public Sector Labour Productivity Number of Public

in Public Sector
Output
YPS
= πPS
* Sector employees
LPS

Since this is an average productivity – the value of the goods and services produced on
average by every public employee – πPS constitutes a synthetic measure of the productivity of
the public sector as a whole, and not only of the labour factor, as in the case of marginal
productivity,10 being in combination with other production inputs. Variations in productivity
per employee can be generated by variations in price or average value of product per
employee, given the physical quantities supplied of each good or service (in which case,
within the framework of the model it will be a matter of variations in the efficacy of the
public administration), or by an increase in the physical quantities per employee, given their
prices and values (variations in efficiency), or by both quantities.
We assume that e-Government processes contribute to GDP growth along three channels:
1. Direct variations of the efficiency and efficacy of the public administration, leading to
increases in labour productivity in the public sector (πPS). Given the number of employees in
the Public Sector (LPS), the consequent increase in value of the public output (YPS) will
translate as growth in GDP (or, it is hypothesised that increases in productivity will be
followed by less than proportional reductions of staff).
2. Direct impact on the private production of goods and services, thanks to multiplier and
accelerator mechanisms connected with the public demand for investment goods and services
(full employment is assumed not to obtain) and the creation of public capital in the form of
material and immaterial infrastructures, assumed at least partially to constitute positive inputs
also for production in the private sector.
3. Direct impact on growth in the private sector generated by the stimulation to innovate, and
by the contribution to the competitiveness of the economic system stemming from the
changed composition of public demand, oriented (in the case of e-Government processes)
towards markedly innovative, high value added goods and services.
The theoretical model we look to refers to the first of these three channels, distinguishing five
mechanisms by means of which variations in efficiency and efficacy lead to variations in

9
See the documents produced within the eGEP Project, for extensive illustration of the model and discussion of
the problems involved in transposing it to the public sector. Cf. Corsi et al. (2006).
10
See Sylos Labini (1995) for a clear exposition of the reasons for this choice and for a theoretical critique of the
neoclassical production function and related empirical applications. Most Sylos Labini’s works are available on-
line at http://dspace.unitus.it/handle/2067/163.

5
productivity – three originally identified by Sylos Labini (1984) – the Smith Effect, the
Ricardo Effect and the Investments Effect (here renamed Schumpeter Effect) – and two more
effects with specific reference to innovation processes in the public administration – the
Back-Office Effect and the Take-Up Effect.
The Smith Effect. At the private level, the Smith Effect connects labour productivity with the
market size of an individual firm (thus an effect defined at the microeconomic level): in
particular, it summarizes the impact of dynamic economies of scale on labour productivity. In
fact, with variations in firm size the efficiency with which the endowment of fixed and
circulating capital is used also varies: generally there will be increasing economies of scale, or
in other words the Smith Effect is expected to exhibit a positive sign, at least through the
possibility of amortising the fixed costs over a larger set of goods and services.
Adapting this concept to the public sector proves far from simple or direct. In fact, in their
activity of supplying services to the community the public administrations have no market (in
the sense of traditional microeconomics) for their products. In many cases there is no
“demand” for public goods, in the sense of evident readiness to pay for them, and production
decisions are guided, rather, by the supply side. At the same time, effectively achieving
improvements in efficiency means launching reorganisation processes that cannot be
considered automatic, given the lack of competitive stimulus of an objective of monetary
gain.
Finally, the reverse relationship, from increases in productivity to increases in scale, which
can be hypothesised in the private sector, thanks to the possibility of setting lower prices at
larger scales on the strength of higher productivity, but it cannot be considered automatic in
the case of the PAs. In fact, the need is for the greater potential supply of public goods and
services to be effectively matched by demand of users and citizens; otherwise, the technically
feasible increases in productivity will remain unfulfilled, unless through staff reduction. At
the aggregate level this reverse effect implies efficacious planning of the broad mix of public
goods and services supplied, meaning by efficacious that it effectively answers to the needs of
the citizens, which may be considered a sort of demand for public services and goods.
In relation to the public sector, the Smith Effect can be broken down into two effect
typologies: microeconomic and macroeconomic. The former applies to the benefits strictly
achieved by the individual public administrations, many of which take the form of gains in
financial terms. In particular, e-Government is able to produce the following intermediate
results: savings in terms of reduction of the cost of services as a whole and/or of single
transactions; reallocation of human and financial resources in favour of those services that are
of the greatest utility to users (increase in efficacy);11 greater integration, customisation and
speed in the supply of goods and services; services supply of new design, and potentially
corresponding new revenues.
At the macroeconomic level, over and above aggregation of the micro-effects, which does not
correspond to their sum it is at least worth noting the increased speed and coverage capacity
of tax revenues.
Indicating with a circumflex accent the rate of variations of the individual variables in period
t, the Smith effect can be indicated in symbols as

11
See Danziger and Viborg Andersen (2002), Grönroos and Ojasalo (2004), Berman and Vasudeva (2005), for
examples of in-depth examination of the efficacy and quality of public services provision.

6
 ∧
π =bYˆSP

where b represents the elasticity of productivity variation to variations in output.

The Ricardo Effect (or Substitution/Integration effect). According to Sylos Labini’s
productivity function model, in the private sector there may or may not be a certain static
substitutability among production inputs, but there certainly is a dynamic substitutability,
associated with process innovation and modification of production technologies. In particular,
in response to variations in the prices of labour and capital goods, firms will be stimulated to
adopt organisational and technological innovations reducing the relative use of the input that
has become more costly.
The public administrations are subjected to far more stringent constraints than are private
firms with regard to the expediency of or needs for variation in the staff employed, especially
when it is a matter of shedding staff. Thus, it seems very likely that variations in prices can
lead to increased efficiency thanks to the integration (and relative increase) of innovation
processes and e-Government with traditional processes, rather than their immediate
substitution.
There are two variables to consider in estimating the Ricardo effect: variation in the average
wages of public employees (w), and the index number of the prices of investment goods
acquired by the public sector ( PI,PS ). Thus, indicating with c the sensitivity of variations in
productivity to variations in relative prices, the variations in productivity can be expressed as
a function of the Smith effect and the Ricardo effect:

∧ ∧

π =bYˆSP +c w ∧
P I ,PS

Before variations in the relative prices lead to the adoption of different technologies there is a
certain time lag, just as the impact of these innovations on productivity will not be immediate:
the effective temporal dimension of the lag is a matter of empirical nature, and we have
therefore omitted time indexes in our general formulation of the model.
The Schumpeter Effect (or effect of investments in innovation). In the last decade, many
studies have been conducted to identify the benefits of ICT investments in terms of
productivity, especially in the private sector.12 Basically, there are two reasons why
investments increase not only the potential output, but also efficiency and/or efficacy in the
supply of goods and services. Firstly, they are sometimes made for this precise purpose; on
the other hand – even if they are made simply to increase the volume of production, thus with
a proportional increase of employees, productivity being equal, or when the aim is to replace
capital either obsolete or old – the introduction of new machinery generally leads to
improvement in operations, thanks to the “embodied” technical progress.
Obviously, not only the mere acquisition of physical goods can be considered investment: in
relation to reorganisation processes and the introduction of forms of e-Government, there are

12
For a review of the economic literature see Brynjolfsson and Hitt (1998); Lehr and Lichtenberg (1999);
Triplett (1999); Dewan and Kraemer (2000); van Ark (2000); Pohjola (2001); Inklaar et al. (2003); van Ark and
Piatkowski (2004).

7
four items to be distinguished: spending on hardware (generally greater in the initial stages of
e-Government processes), spending on software (also greater at the stage of introduction, but
fairly steady in the subsequent stages), spending on external consulting and on staff training
plans (greater at the more advanced stages of the innovation process).13
Indicating with I expenditures on investment in the public sector, and again adopting the
convention of ignoring the temporal dimension of the individual effects, we obtain the
following productivity function:

∧ ∧

π =bYˆSP +c w ∧ +d I ,
P I ,PS

where d represents the elasticity of variations in productivity generated by public investments.

The Back-Office Effect and the Take-Up Effect. Transposing the productivity function model
to the public sector entails at least two other types of considerations. The two effects
introduced by Corsi et al. (2006) are multiplicative in relation to the previous effects, in that
they afford greater or lesser effectiveness to the dynamics so far defined. In any case,
identifying them empirically proves all too formidable a task, given the lack of much of the
relevant data and the considerable difficulties involved in measuring the interest variables: the
two effects will not be taken into account in the following econometric analysis.
The Back-office effect includes the impacts on reorganisation processes induced by ICT
implementation initiatives, taking into consideration the potentially greater rigidity of the
public sector towards modernisation phenomena in comparison with the private sector.14 In
order to completely achieve the benefits of e-Government in terms of efficiency and
effectiveness, PAs are obliged to accomplish high levels of integration among their various
organisational areas and units. Re-engineering the back-office functions is a primary factor to
be considered for the creation of an e-Government structure able to provide integrated and
efficient public services, and it appears all the more important when we recall the minimal
impact of the first e-Government projects carried out up to a few years ago. As they focused
solely on a simple and rapid translation online of the very same traditional public services,
without any concrete reorganisation of the productive processes, they mostly failed.
The Take-Up effect can be defined as a set of environmental conditions that enable e-
Government implementation and determine its efficacy. Here there are at least two significant
aspects to distinguish:
- The technological scenario: one may assume that the demand of citizens for public services
exploiting ICT increases with the supply of private ICT – related services and products.
- The competition of private services could drive the public sector to greater efforts to achieve
a more rapid and efficient supply of services.
- The educational/training level of the staff employed in the public sector and of the entire
population is decisive for the supply of knowledge-based services.

13
Although - strictly speaking - expenditure borne indirectly on the reorganisation of processes and services
should also be calculated among the investment in innovation cost items, rating them is empirically far harder,
and they will therefore be omitted from the following analysis.
14
Cf. Bertschek and Kaiser (2004).

8
Broadly speaking, the social environment influences the efficacy of e-Government
programmes – or in other words their impact on the productivity of the public sector – on both
the demand side, with greater receptivity of the potential users, and on the supply side, with
better-prepared staff in the public offices.
Indicating with ϕ the set of context variables affecting the efficiency of the effects considered,
and with ψ the capacity of the policy-makers to reorganise the public sector as a whole in
response to the incentives considered, thus we may sum up the productivity function:

∧ ∧

π =(ϕ,ψ ) bYˆSP +c w ∧ +d I
P I , PS

4. ICT, efficiency and efficacy in the public sector

The development of the Information Society, and in particular the increasing use of ICT as a
channel for interaction between citizens or firms and the public administrations, has spawned
a great many studies, seeking to size up the potential and key areas of impact of e-
Government, thus offering policymakers a better understanding of its benefits in terms of
efficiency and efficacy. It is worth recalling here some of the most significant studies carried
out on the subject of e-Government.
With regard to the efficacy of public services, the sample survey carried out annually by
Capgemini since 2001 represents a major contribution.15 Taking a methodological approach
of the benchmarking type, the study analyses twenty basic public e-Services, supplied by the
public administrations of Europe to citizens and firms, on the basis of two main indicators:
full online availability (in the sense of the number of services that can be fully provided on
electronic platforms) and level of online sophistication of the service (in particular, five levels
are distinguished, from non-availability of online service to a stage at which the procedures
forming the service are characterised by perfect integration between PA and user). The
findings of the last available report show significant advances in both indicators: in the 27 EU
countries, on average nearly 50% of the services are available through the Internet, in
comparison with the 40% of the year before, with a fairly high level of sophistication (level
4).
Also based on a benchmarking methodology is UNDERSTAND,16 a project promoted and
coordinated by the Regione Emilia Romagna, concluded in 2006, the aim of which is to
compare the degree of development of the Information Society (and in particular e-
Government) at the regional level in Europe, defining and applying a set of common
indicators. The long-period results expected are substantiated in supplying the twelve regions
involved in the project with instruments to evaluate the impact in terms of efficacy of
investments in ICT, in the ambits of both the public administrations and the social
environment in general (firms, schools, citizens, etc.). The surveys carried out confirm
advance in the innovation processes of the regional PAs, but at the same time reveal certain
limits. As the complexity of the services increases, their availability electronically decreases;
multi-channel supply remains at the primordial stage; and efforts must be made to reduce the

15
Capgemini (2006).
16
The documents produced in the course of the project are available on the website www.understand-eu.net. For
a summary of the results see Mancini (2006).

9
negative relationship there is between PA dimension and the development of e-Government
initiatives.
With regard to evaluation of the efficiency and efficacy of public services, it is worth citing a
work published in 2004 by Cisco Systems,17 containing interviews with over 1400 people
responsible for investment choices in relation to the supply of electronic public services (at
both the technological and organisational level) working in the central, regional and local
public administrations of eight European countries. The study determines a series of factors of
critical importance in achieving increases in the efficiency of e-services supply (for example,
the average time taken to complete a procedure, average cost of a procedure, total number of
procedures concluded within a given span of time, etc.), so as to identify, on the basis of these
key aspects a series of best practices to imitate.
It hardly needs pointing out that much of the literature on efficiency and efficacy in the public
administrations is of the organisational-management type and based on sample surveys of best
practices, often characterised by considerable use of methodologies of the benchmarking type.
The eGEP Economic Model (Corsi et al., 2006) marks a break in the line of the literature, in
that it addresses the complexity of the subject systematically (i.e. analysing the entire public
sector) while strictly grounded on economic theory.
The points developed there also underlie the model presented here, the subject of which – as
we have seen – is evaluation of the efficiency shown by the public administrations in the
supply of goods and services, which are assumed to be comparable on the basis of market or
estimated value. It will also emerge from the analysis that this approach affords some general
indications for a preliminary evaluation of the efficacy of the PAs. The reason for researching
aspects associated with efficiency lies mainly in the lesser complexity of an empirical survey,
thanks to the (relatively) greater measurability of the variables involved.
A final point to note is the new viewpoint taken – pre-analytic, in the terminology of
Schumpeter (1954) – which makes comparison of the model here proposed with the economic
and organisational literature on PA somewhat difficult, in that the main assumption adopted
sees increasing efficiency in the PA as precondition for the supply of more and better
products and services, rather than intermediate objective on the way to downsizing the role of
the public sector in the economy. Actually, the model presented here also takes account of the
issue of reducing the bureaucracy weighing on firms and citizens, but it goes further, seeking
to determine how the public sector can, on the strength of innovations guided by ICT
implementation, actively enhance its own capacity and generate a positive impact on
economic growth. Here the key assumption is that the objective of arriving at public
administrations able to supply services conceived in terms of the users’ needs, can in the first
place be achieved by boosting productivity through reorganisation, professional training and
ICT, or in other words through e-Government.

5. Public sector productivity dynamics in some OECD countries

For the reasons mentioned in section 2, constructing a public sector productivity database is
not an easy task. However, even with the inevitable limits involved in the necessary
approximations of the theoretical variables, some interesting results may be achieved by
considering the OECD countries. As noted in the mentioned Atkinson Report, these countries
exhibit considerable improvements in the recent years, concerning the quality of data on
public sector production, especially in terms of the share of public sector output measured

17
Cisco Systems-Momentum (2004).

10
directly or through some indirect measures that do not exclusively or mechanically rely on the
value of inputs. Thus, especially the data concerning the most recent years may in a first
approximation be employed to highlight the potential for improved productivity in the public
sector.
The database considered includes 24 OECD countries18 observed from 1998 to 2005: we do
not consider the most recent years at this stage because of lack of updated data on ICT-related
expenditure and in order to avoid the strong perturbation effects due to the financial and
economic crisis and the following public austerity programs.
Public sector output is proxied with the value of the production of the Central, Regional and
Local Administrations (since institutional differences do not affect the estimates); the same
productive units are considered for the dynamics of the average wages and the number of full-
time equivalent employees in the public sector (i.e. adjusted for the hours worked).
Investments are expressed in the form of gross fixed capital, that is before accrued
depreciation. Direct ICT expenditure, i.e. excluding the expense incurred in reorganising
production, is provided by the WITSA database.19
The formulation adopted considers rates of change, with regard to both labour productivity
and the Smith and Ricardo effects, for the sake of better international comparability of the
data, as compared to e.g. first differences, given the partially different accounting standards
across countries. The variation of public output (Smith Effect) is taken with a one-year time
lag to avoid possible spurious correlations with the productivity dynamics, given the possible
rigidity of employment. The Ricardo Effect is also considered with a lag for theoretical
reasons. Public investments are expressed as percentages of public output, and e-Government
expenditure as a percentage of GDP: the former with one lag, the latter with two. The model
is firstly estimated in its basic form, then including e-Government. Productivity growth is
considered on both a yearly and a long-period basis. Pooled estimates are conducting jointly
considering variation across all countries and all years at the same time, with proper
corrections of the standard errors considering correlation across observations regarding each
country and heteroskedacity of errors across countries.
Considering the model in the simplest form (Table 1, columns denoted as OLS), it emerges
that the productivity function accounts for 52% of productivity growth on an annual basis,
and for an eve greater proportion of variance in the case of long-run growth, up to 83% on a
triennial basis. The significance of the Ricardo Effect and of investments increases with the
increase in the time span considered, although the sample size is correspondingly reduced
(implying a reduction in the width of the confidence intervals). Both observations clearly
indicate that the model is better equipped to capture productivity growth determinants in the
medium-to-long run than in the short run.

Table 1. Productivity growth in the public sector, 2000-2005

Annual var. Biennial var. Triennial var. Quadrennial var.

OLS IV OLS IV OLS IV OLS IV

18
Australia, Austria, Belgium, Canada, the Czech Republic, Denmark, Finland, France, Germany, Greece,
Holland, Hungary, Ireland, Italy, Mexico, Norway, New Zealand, Poland, Portugal, Slovakia, Spain, Sweden,
Switzerland, Turkey.
19
WITSA (2006).

11
 Smith 0.239 0.262 0.034 -0.047 -0.169 -0.047 -0.222 -0.159
(0.164) (0.112)* (0.084) (0.134) (0.197) (0.228) (0.340) (0.346)

Ricardo 0.178 0.201 0.468 0.512 1.131 1.046 0.999 1.005

(0.178) (0.124) (0.138)** (0.135)** (0.230)** (0.250)** (0.204)** (0.188)**

Investments -0.016 -0.056 0.285 0.068 0.957 0.131 1.465 0.776

(0.081) (0.094) (0.174) (0.168) (0.285)** (0.196) (0.457)** (0.824)

E-Government 0.975 8.738 12.815 9.803

(1.368) (4.766) (4.912)* (13.818)

Observations 158 93 115 91 69 65 25 25

R2 0.52 0.68 0.83 0.77
Notes: Standard errors robust to heteroskedacity and self-correlation in brackets. The estimate includes annual
dummies as control variables.
* significance 5%; ** significance 1%

With more prolonged time spans, the coefficients of the two variables (Ricardo and
Investments) increase, showing a positive role for public investments in the medium run – not
observed in the short run – and an efficient response to market signals, despite the lack of
competition in the public supply of many goods and services. It is to be noted that all the
coefficients show marked variance, indicative of a certain heterogeneity at the national and, in
some cases, temporal level. In particular, the high variance causes the Smith Effect to be not
significant at the traditional confidence levels. Furthermore, the respective coefficient
decreases on average, and in the minimum and maximum, as longer time spans are
considered. The conclusion this brings us to, is that the public administrations are unable to
achieve economies of scale when they grow in size. Behind this incapacity there may lie a
technological impossibility (associated with the technical conditions of goods and services
supply), or inability to reorganise the overall organisation with due efficiency, or it may
depend on the typology of the goods and services supplied efficiently (i.e. effectively
demanded by the citizens). It is also to be pointed out, however, that the Smith Effect has a
substantially variable influence on productivity over time, as it emerges from the separate
estimation of cross-sections for the various years (shown in the Appendix): in the sample of
countries considered a positive trend emerges – albeit more notably in short-period growth
than over the medium-long term – thanks to which the Smith Effect proves significantly
positive in most of the formulations referring to years 2004 and 2005.
Introduction of e-Government expenditure in the model considered, appears empirically
problematic given the high correlation between e-Government expenditure and public
spending on investments, despite the different normalisation of the two magnitudes. Thus
estimations were made, pooled and individually for each year, with the instrumental variable
method (IV columns in Table 1). In fact, the database considered offers details of the
distribution of ICT expenditure in the four categories of software, hardware, services and
consultation, and communication. These variables, expressed as percentages of national ICT
expenditure, together with the quota of public expenditure for ICT over the total national ICT
expenditure, are closely correlated with e-Government spending, but not with productivity
dynamics, nor with public investments as a whole (see Table A2 in the Appendix): they thus
prove excellent instruments to estimate the impact of e-Government on productivity.

12
In all the specifications, estimation of e-Government expenditure in relation to the five
instruments described accounts for 88% to 98% of the variance; on the other hand, R2 does
not prove a significant measure in relation to the second stage of estimation. Explicit
consideration of e-Government implies certain modifications to the preceding results. While
the Smith Effect becomes significantly positive in estimation of the annual variation in
productivity, the Ricardo Effect shows greater relevance over the medium-to-long run,
growing in magnitude and significance. At the same time, the Investments coefficient
diminishes as the period under consideration is prolonged, without a corresponding reduction
in variance (which, moreover, increases in some cases). This tends to downsize the role of
public investments in the increase of public sector productivity, in favour of expenditure
specifically going into e-Government programmes, which likewise loom gradually larger.
This finding is hardly surprising when we consider that, unlike e-Government programmes,
public investments are not necessarily channelled into enhancing the productivity of public
employment (as in the case of building infrastructures, for example, or when directed to an
increase in productive capacity and not in productivity). Nevertheless, the fact that their
contribution to the growth of productivity in the public sector appears not only modest but
statistically insignificant, signals the need for further research in analysing the efficacy of
public investments.

6. Implications for Economic Policy

The analysis presented in this work shows that the public sector can fruitfully be given a
representation in terms of the sum total of several public productive units. A vision of public
administrations as a productive sector of the economy can be given a meaningful quantitative
representation through a theoretical model and an empirical estimation not totally different
from those applicable to the private sector. Truly, the social value of political and economic
public interventions should be understood as wider and possibly more important than simply a
production-increasing effort, but still we claim that conceiving of the public sector as a
productive unit may help us providing an economic rationale for it.
For example, we show that investments in the enhancement of public sector efficiency and
efficacy do bring positive economic outcomes. The reorganisation of the public
administrations prompted by the adoption of ICT promises to increase the production and
diffusion of information, making the public administrations more transparent and responsible
towards the citizens and policy-makers, reducing the scope for corruption and enhancing
opportunities for all citizens and firms. From a strictly economic point of view, no less
potential is shown by the e-Government initiatives: indeed, e-Government represents an
opportunity for radical transformation of the PAs, both in terms of goods and services
supplied to the citizens and of capacity to satisfy needs not adequately met by the market or
the informal sector (public sector efficacy), as well as in terms of efficiency in the supply of
these services and in support for the services supplied by the market and the family (where the
public sector constitutes a factor of production).
Focussing on the impact of e-Government on public sector productivity and defining the
results in terms of GDP growth, our analysis appears to yield some encouraging preliminary
findings. In many OECD countries investments in ICT have contributed positively on
productivity growth in the public sector, more effectively and significantly than have public
investments as a whole. As one would have expected, both forms of investment show more
importance in the dynamics of the medium-to-long run than on the short run.
Contrary to expectations voiced from time to time, the public administrations seem to respond
efficiently to market incentives in terms of the prices of inputs, especially over the long
13
period. Moreover, although the public sector is subject to dynamics that often do not apply to
the private sector, for example in the field of staff management, the average productivity of
public employees appears to an appreciable extent to be determined by the dynamics of the
capitalistic sector of the economy. This is an interesting finding since it suggests substantial
dynamic efficiency in the public sector, pointing to the expediency of further empirical
investigation in the face of repeated calls for the minimisation of public intervention in the
economy on the grounds of its alleged inefficiency.
Finally, our analysis highlights the need for further research in the field of public sector
efficacy: since increases in output are not significantly correlated with increases in
productivity (apart from the short period), specific inquiry needs to be made into the extent to
which this limit may be due to the production technology of the public sector – which would
rule out economies of scale – rather than to incapacity to reorganise production at the micro
(Back-Office) level, or the macroeconomic level (in the composition and nature of the goods
and services supplied).

References
Berman O. and Vasudeva S. (2005), “Approximating Performance Measures for Public Services”,
IEEE Transactions on Systems, Man, and Cybernetics - Part A: Systems and Humans, vol.35, n.4,
pp.583-591.
Bertschek I. and Kaiser U. (2004), “Productivity Effects of Organizational Change: Microeconometric
Evidence”, Management Science,vol. 50, n. 3, pp. 394-404.
Brynjolfsson E. and Hitt L.M. (1998), “Beyond the Productivity Paradox”, Communications of the
ACM, vol.41, n.8, pp.49-55.
Capgemini (2006), Online Availability of Public Services: How Is Europe Progressing?, Web Based
Survey of Electronic Public Services, Report of the 6th Measurement.
Cisco Systems – Momentum (2004), Net Impact: From Connectivity to Productivity.
Corsi M. and Guarini G. (2007), “La fonction de productivité de Sylos Labini: aspects théoriques et
empiriques”, Revue d'économie industrielle, n.118, pp.55-78.
Corsi M., Gumina A., D’Ippoliti C. and Battisti M. (2006), eGEP Economic Model, Final Report on
the Benefits, Costs and Financing of eGovernment, delivered for the European Commission, DG
Information Society, Modinis Programme.
Cutler D. M. and Berndt E. R. (eds.) (2001), Medical Care Output and Productivity, University of
Chicago Press.
Cutler D. M. and McClellan M. (2001), “Productivity Change in Health Care”, The American
Economic Review , vol. 91, n. 2, Papers and Proceedings of the Hundred Thirteenth Annual Meeting
of the American Economic Association, pp. 281-286.
D’Ippoliti C. and Roncaglia A. (2011), “L’Italia: una crisi nella crisi”, Moneta e Credito, vol. 64, n.
255, pp. 189-257.
Danziger J. N. and Viborg Andersen K. (2002), “The Impacts of Information Technology on Public
Administration: An Analysis of Empirical Research from the ‘Golden Age’ of Transformation”,
International Journal of Public Administration, vol. 25, n.5, pp. 591 - 627.
Dawson D., Gravelle H., O’Mahony M., Street A., Weale M., Castelli A., Jacobs R., Kind P.,
Loveridge P., Martin S., Stevens P. A. and Stokes L. (2005), “Developing New Approaches to
Measuring NHS Outputs and Productivity”, NIESR Discussion Paper n. 264 and CHE Research Paper
n. 6.

14
Dewan S. and Kraemer K.L. (2000), “Information Technology and Productivity: Evidence from
Country-Level Data”, Management Science, vol. 46, n. 4, pp. 548 - 562.
Giordano R. and Tommasino P. (2011), “Public Sector Efficiency and Political Culture”, Temi di
discussione – Working Papers, n. 786, Bank of Italy, Rome.
Glassner V. (2010), “The Public Sector in the Crisis”, ETUI Working Paper, n. 2010.07, European
Trade Union Institute, Brussels.
Grönroos C. and Ojasalo K. (2004), “Service Productivity: Towards a Conceptualization of the
Transformation of Inputs into Economic Results in Services”, Journal of Business Research, vol. 57,
n. 4, pp. 414 - 423.
Inklaar R., O’Mahony M. and Timmer M.P. (2003), ICT and Europe’s Productivity Performance:
Industry-level Growth Account Comparisons with the United States, GGDC Research Memorandum
GD-68, Groningen Growth and Development Centre, December.
Kregel J. (2011), “Resolving the US Financial Crisis: Politics Dominates Economics in the New
Political Economy”, PSL Quarterly Review, vol. 64 n. 256, pp. 23-37.
Lehr B. and Lichtenberg F. (1999), “Information Technology and Its Impact on Productivity: Firm-
Level Evidence from Government and Private Data Sources, 1977-1993”, The Canadian Journal of
Economics, vol. 32, n. 2, Special Issue on Service Sector Productivity and the Productivity Paradox,
pp. 335 - 362.
Mancini C. (2006), “Un ruolo da traino nelle regioni europee”, Il Sole 24 Ore – Nòva, 27 aprile, p.6.
NSO (2005), “Measurement of Government Output and Productivity for the National Accounts”,
Atkinson Review: Final Report, Palgrave MacMillan, London.
O’Mahony M. and Stevens P. A. (2003), International Comparisons of Performance in the Provision
of Public Services, National Institute of Economic and Social Research, London.
Pohjola M. (ed.) (2001), Information Technology, Productivity, and Economic Growth, Oxford
University Press, Oxford
Sylos Labini P. (1984), Le forze dello sviluppo e del declino, Laterza, Roma-Bari.
Sylos Labini P. (1985), The Forces of Growth and Decline, MIT Press, Cambridge (MA).
Sylos Labini P. (1995), “Why the Interpretation of the Cobb Douglas Production Function Must Be
Radically Changed”, Structural Change and Economic Dynamics, vol.6, n.3, pp. 485-504.
Sylos Labini P. (2004), Torniamo ai classici, Laterza, Roma.
Schumpeter J.A. (1954), History of Economic Analysis, George Allen & Unwin, London.
Stevens P.A., Stokes L. and O’Mahony M. (2006), “Metrics, Targets and Performance”, National
Institute Economic Review, vol. 197, n. 1, pp. 80-92.
Stiglitz J.E., Sen A. and Fitoussi J.-P. (2011), Report by the Commission on the Measurement of
Economic Performance and Social Progress, available online at the URL: http://www.stiglitz-sen-
fitoussi.fr/documents/rapport_anglais.pdf
Triplett J. E. (1999), “The Solow Productivity Paradox: What do Computers do to Productivity?”, The
Canadian Journal of Economics, vol. 32, n. 2, Special Issue on Service Sector Productivity and the
Productivity Paradox, pp. 309 - 334.
van Ark B. (2000), “Measuring Productivity in the ‘New Economy’: Towards a European
Perspective”, De Economist, n. 148, pp. 87 - 105.
van Ark B. and Piatkowski M. (2004), Productivity, Innovation and ICT in Old and New Europe,
Research Memorandum GD-69, Groningen Growth and Development Center, March.
WITSA (2006), Digital Plane 2006: The Global Information Economy, WITSA.

15
 APPENDIX

Table A1. Cross-section estimates of productivity variance

Annual Variation

2002 2003 2004 2005

OLS IV OLS IV OLS IV OLS IV
Smith 0.046 0.210 0.540 0.189 0.370 0.414 0.640 0.754
(0.276) (0.330) (0.315) (0.282) (0.124)** (0.201) (0.175)** (0.234)**

Ricardo 0.572 0.355 0.102 0.327 0.078 0.036 -1.235 -1.312

(0.396) (0.434) (0.330) (0.278) (0.298) (0.370) (0.535)* (0.580)*

Investments 0.362 0.028 0.650 0.033 0.190 0.211 -0.248 -0.228

(0.102)** (0.151) (0.186)** (0.158) (0.113) (0.108) (0.177) (0.203)

eGovernment 6.867 13.461 -1.147 -1.663

(3.599) (3.071)** (3.219) (1.838)

Obs 24 24 24 24 24 24 20 20
R-sq 0.52 0.75 0.80 0.28

Biennial Variation

2002 2003 2004 2005

OLS IV OLS IV OLS IV OLS IV
Smith -0.510 -0.323 -1.081 -0.663 -0.209 -0.157 0.308 0.331
(0.390) (0.358) (0.347)** (0.405) (0.242) (0.223) (0.071)** (0.064)**

Ricardo 1.100 1.104 0.902 0.740 0.946 0.814 -0.728 -0.750

(0.531) (0.530) (0.190)** (0.216)** (0.538) (0.441) (0.493) (0.503)

Investments 0.289 0.011 0.696 0.170 1.407 0.296 0.262 0.308

(0.201) (0.221) (0.257)* (0.233) (0.289)** (0.222) (0.181) (0.213)

eGovernment 6.469 13.831 21.034 -1.468

(6.219) (6.693) (4.646)** (2.742)

Obs 22 22 24 24 24 24 20 20
R-sq 0.47 0.82 0.75 0.70

Triennial Variation

2002 2003 2004 2005

n.d. n.d. OLS IV OLS IV OLS IV
Smith -0.375 0.015 -0.826 -0.454 -0.080 0.188

16
 (0.483) (0.439) (0.414) (0.351) (0.336) (0.337)

Ricardo 1.318 1.168 1.610 1.415 1.038 0.704

(0.304)** (0.318)** (0.325)** (0.290)** (0.540) (0.508)

Investments 0.915 -0.073 1.073 0.310 1.450 0.349

(0.349)* (0.285) (0.357)** (0.281) (0.403)** (0.298)

eGovernment 19.194 18.778 21.291

(7.329)* (5.088)** (6.527)**

Obs 17 17 22 22 20 20
R-sq 0.82 0.89 0.81
Note: Standard errors robust to heteroskedacity in brackets. * significance 5%; ** significance 1%

Table A2. Simple correlations: e-Government, public investments, productivity of the public
sector, instruments adopted

Hardware Software Services Communic. Public

Share Share Share Share Share
E-Government Expenditure -20,89% 57,89% 69,27% -56,98% 78%
Investments -18,38% -4,64% -11,54% 13,83% 12,12%
Productivity annual var. 2,84% 9,32% 0,57% -3,65% -3,25%
Productivity biennial var. -10,34% -12,69% -0,55% -0,23% -4,55%
Productivity triennial var. -23,54% 10,44% -6,03% 7,66% -9,7%
-
Productivity quadrennial var. -22,62% 2,42% -15,51% 16,01% 20,72%

Table A3. e-Government expenditure estimates in relation to the instruments

Annual Biennial Triennial Quadrennial

Expenditure Expenditure Expenditure Expenditure
Hardware Share -0.0402 -0.0409 -0.0329 -0.0189
(0.0081) (0.0081) (0.0103) (0.0249)

Software Share 0.0176 0.0194 -0.0004 -0.0179

(0.0142) (0.0139) (0.0176) (0.0309)

Services Share 0.0255 0.0259 0.0302 0.0292

(0.0071) (0.0069) (0.0088) (0.0191)

Communications Share 0.0072 0.0089 (0.0087) 0.0034

(0.0019) (0.002) (0.0027) (0.0054)

Public Sector Share 0.0388 0.0377 0.0347 0.0438

17
 (0.008) (0.0079) (0.0098) (0.0203)

Smith 0.0004 -0.0018 -0.005 -0.0034

(0.0036) (0.0018) (0.0021) (0.0027)

Ricardo 0.0007 5.59e-06 0.0011 0.0004

(0.0042) (0.0017) (0.0021) (0.0023)

Investments -0.0044 -0.0051 -0.0033 0.0023

(0.0035) (0.0035) (0.0046) (0.0113)

Observations 93 91 65 25
R2 96,75% 96,94% 96,88% 97,13%
Notes: Standard errors robust to heteroskedacity and self-correlation in brackets. The estimate includes
annual dummies as control variables.

18