Download Outsourcing and efficiency

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Privatization wikipedia , lookup

Transcript
The unattainable efficiency - interpreting apparent inconsistencies in outsourcing
and privatization
Abstract
Cost-containment coupled with a multidimensional constraint on quantity, quality and equity
encourages policies promoting efficiency in the production of public services. Managing the
boundaries of direct production and of public ownership have been two of the most popular
policies for designing public services since the diffusion of New Public Management (NPM)
theories. However the impact of outsourcing and privatization on efficiency is not clear and
policymakers seem to follow chaotic strategies. In this paper we try to shed new light on the
relationship between outsourcing, privatization and public expenditure and give an
interpretation to the apparently inconsistent behavior of policymakers. Using data from the
Italian National Health Service we map Regions according to their choices about the
outsourcing/privatization mix, both from a static and a dynamic point of view. We explain our
results using Transaction Costs Economics, Incrementalist theory, Punctuated Equilibria and
Garbage Cans.
Keywords: Privatization, Health Care, Quantitative
Over the last few decades, the necessity of reducing the gap between available financial
resources and public expenditure has become one of the most discussed issues in the public
sector, and increasing attention has been given to strategies for controlling input costs with a
multi-dimensional constraint on the output (Buchanan, Cappelleri, & Ohsfeklt, 1991; Burke &
Goddard, 1990; Hodges, 1997; Murray, Erridge, & Rimmer, 2012). Consequently, efficiency
has become an important objective for public organizations. The debate has thus far been
articulated over the following two main logical levels: i) the root question about the definition
1
of what are (and should be) the boundaries of public services (Hefetz & Warner, 2012; Kettl,
2006; Levin & Tadelis, 2010; Mailly, 1986) and ii) second-order concerns about how to
efficiently organize and control service provisions and financing once the boundaries have
been decided. Instances of contributions that are classified into this second category are
generally focused on one specific domain of public services, such as transportation (Zullo,
2008), services delivered by municipalities (Anessi-Pessina & Steccolini, 2005; Kettl, 2006;
Levin & Tadelis, 2010) and, perhaps more than any other, healthcare (Armeni, 2011;
Buchanan, Cappelleri, & Ohsfeklt, 1991; Mapelli & Lucioni, 2003; Vernon, 2002).
Scholars have studied these questions using three main research approaches: positivistic,
interpretive and descriptive. Studies adopting a positivistic perspective aim at identifying a
causal relationship between policies, strategies and some variables of system performance
(Ashenfelier & Card, 1985; Brekke, Holmas, & Straume, 2011; Ferrario & Zanardi, 2011).
The methods generally include econometric analyses or economic modeling. Studies applying
a more interpretive approach try to explain, using qualitative methods, either the success or
the negative unpredicted outcome of public cost-containment actions (Anessi-Pessina &
Steccolini, 2005; Bloching, Stock, & Scheel, 2008; Chalkley & Malcomson, 1998; Enthoven,
2004; Ferrè, Cuccurullo, & Lega, 2012; Murray, Erridge, & Rimmer, 2012). Finally, a
relevant number of studies have focused on the description and analysis of public policies,
with the objective of critically assessing their potential, using previous evidence or policy
impact analyses from a mixed positive/normative perspective (Anessi-Pessina, Cantù, &
Jommi, 2004; Battaglio Jr. & Ledvinka, 2009; Boyne, 1998; Burke & Goddard, 1990).
In this paper, we take a mixed positivistic-interpretive approach and apply it to the case of the
Italian National Health Service (INHS) to investigate to what extent cost-containment policies
at the Regional level gain effectiveness from make-or-buy decisions and from managing
public-private boundaries. We argue that these two types of decisions are not coincident but
2
rather are complementary in effectiveness; however, the transition to a more efficient
configuration is not as straightforward as economic analysis would suggest. Building on
incrementalist theory (Lindblom, 1959; Lindblom, 1979), punctuated equilibria (Breunig &
Koski, 2012) and previous empirical research, we show that the road to efficiency, defined on
the basis of the potential level of affordable savings given the status quo, is inherently
implausible. Therefore, we address the illusion of economic efficiency. In such a scenario,
Regions seem to change strategy without following a clear pattern. In fact, attaining the
potential maximum efficiency would require changes to a rigid institutional framework
(Regional healthcare model) that, once modified, would suggest a different optimal allocation
of make-or-buy decisions and a different boundary between the public and the private.
We contribute to the debate on the effectiveness of cost-containment policies by showing that
the reason for the existing lack of clear-cut results in favor of either make-or-buy or publicprivate decisions lies in the illusion of efficiency, i.e., the mistake of believing that a potential
optimal efficiency is affordable. In doing so, we undertake an empirical efficiency study by
comparing organizational configurations in the Italian healthcare sector and with a simulation
exercise that explains the apparently inconsistent Regional policies over time. Moreover, we
open the question about the nature of this (apparently) infinite process: is it a long way toward
a perpetually changing target through an iterative learning process, consistent with
incrementalism, or is it a random process with erratic new targets, consistent with a garbagecan-like disjunction of problems and solutions (Cohen, March, & Olsen, 1972)?
The paper is structured as follows: in the first section, we present our theory, and we also
introduce the research context of the INHS. In the methods session, we present the statistical
model and the interpretive structure of our analysis. The results and discussion section draws
on empirical evidence and our simulation to explain the puzzle of Regional strategies.
Incrementalist theory and the concept of punctuated equilibria are used to support our thesis
3
about the unattainable nature of efficiency. The concluding remarks highlight drawbacks and
future directions for research.
THEORY AND HYPOTHESES
Efficiency in public service production and delivery is of paramount importance for New
Public Management (NPM)1. According to Lynn (1998), the diffusion of NPM has brought
some constructive legacies, where a prominent role is given to “a stronger emphasis on
performance-motivated administration and inclusion in the administrative canon of
performance-oriented institutional arrangements, structural forms and managerial doctrines
fitted to particular contexts” (Lynn Jr., 1998). Performance is a multidimensional concept.
This concept is particularly evident in the public sector, where the prevalent performance
dimension of private companies – profit – is inherently inappropriate. Because of this
inappropriateness, a broader approach to performance identification and measurement is
preferred (Holzer & Yang, 2004; Newcomer, 2007; Nicholson Crotty, Theobald, & Nicholson
Crotty, 2006; Pollitt, 2006).
Since the late ‘70s, NPM and the tightening of budgets have gained advocacy in developed
countries. Therefore, cost-containment has become one of the most relevant and urgent issues.
Downward pressure on costs implies that there is modification of the objective function of
policymakers in terms of efficiency, with constraint on other performance dimensions (e.g.,
quality and equity).
1
NPM is one of the most debated set of cultural elements and managerial tools. It has been
widely criticized because of its lack of adherence to the core peculiarities of public
management (Riccucci, 2001).
4
In this paper, we focus on outsourcing (make-or-buy) and privatization, two efficiencyimprovement policies that can be taken as an example of quasi-markets introduced under
NPM influence. We chose the case of the Italian Healthcare Service for the possibility of
disentangling these two policies, which are generally viewed as coincident (David & Chiang,
2009; Guerrero, Mossé, & Rogers, 2009), because INHS has interesting variability in the mix
of these policies both across Regions and over time.
Outsourcing and efficiency
In the presence of minimum quality constraints, a decline in spending implies an
improvement in efficiency (Boyne, 1998). Spending, in turn, can be controlled with make-orbuy decisions concerning the separation of two or more activities in a production process with
the objective of exploiting specialization economies (buy) (Levin & Tadelis, 2010) or
reducing transaction costs (make) (Williamson, 1973; Williamson, 1979).
Williamson argued that markets and bureaucracies are two alternative opportunities for
organizing the production structure. In his view, the market is the most efficient
organizational form unless the transactions costs in the market are higher than the transactions
costs incurred by operating within a bureaucracy. The activities are, therefore, allocated to the
most appropriate executor and are defined according to the efficient boundary between
hierarchy and market. Under this perspective, outsourcing in the public sector has often been
associated with the decision for privatization.
However, in some cases, these two have been considered as separate choices, particularly
when: i) the hidden costs of competition in terms of an inefficient market structure or high
prices are manifest and ii) the level of market failures in the sector (e.g., education or
healthcare) is considered to be inappropriate for leaving the production to private actors only
5
(Borgonovi, Fattore, & Longo, 2012). Under such circumstances, some studies have reported
that contracting out to other public agencies2 appears to be largely a substitute for in-house
provision rather than an analogue of privatization (Levin & Tadelis, 2010). Therefore, even in
the absence of privatization, any change in the level of outsourcing in the public sector
produces a variation in the following two categories of costs: transaction costs and
organizational costs. Based on this premise, we expect that the direction of the impact of
outsourcing on spending varies according to the sum of the transaction costs (and specular
benefits of direct control) and the organizational costs (deriving from the management of
complexity).
However, the interplay between these two categories of costs may depend upon the initial
condition as both contracts and organizations have fixed institutional costs3. At different
levels of outsourcing, the decision to expand or reduce the organizational boundaries may
have different impacts on spending, suggesting a non-linear relationship between make-orbuy and public spending. In a study focused on motor bus services (Zullo, 2008), outsourcing
has revealed a weak curvilinear association with the total cost, suggesting that the most costefficient agencies either fully contract out or provide full in-house service. Therefore, we
expect that given the importance of the fixed costs of moving from the extremes, intermediate
configurations are the more costly, identifying an inverted U-shaped relationship between
outsourcing and spending.
2
The diffusion of NPM-style reforms have often produced the disintegration of vertically
integrated bureaucracies substituted with arm’s length executive agencies.
3
For example, if the initial condition is a pure market configuration, then insourcing back
some activities may entail substantial fixed costs for acquiring production factors that may
overrun the savings in the transaction costs. Similarly, from a pure bureaucracy configuration,
outsourcing-related costs may include relevant transaction costs for beginning to interact with
other institutional actors.
6
In those INHS Regions where outsourcing is set at an intermediate level, the sum of the
transaction costs and organizational costs is high, suggesting an opportunity for reducing
public expenditure via make-or-buy.
Hypothesis 1. In the absence of privatization, outsourcing has an inverted U-shaped
relationship with public expenditure.
Privatization and efficiency
Privatization is the other decisional dimension for macro-organizing public services4. The
motivation for privatization is rooted in the intended benefits of competition as a mediator for
efficiency5. For example, Pendleton (1999) examined whether privatization and exposure to
product market competition (in local transportation) leads to changes in labor management
and industrial relations, leading to relatively higher levels of efficiency (Pendleton, 1999).
However, competition is not granted by the presence of some private actor in the production
system. For example, competition was not recognized as widely diffused in U.S. healthcare,
even though more than half of the care provision is privately managed and financed. Notably,
health care systems seem to be characterized by status/quality competition (Ferlie, 2007)
rather than by price or public/private competition. Levin and Tadelis (2010) found that
privatization had a significant impact in improving efficiency only if the conditions exist for a
4
In the particular case of the INHS, the configuration where the public sector does not
contract out excludes the presence of private providers of care as emerging entrants (model 3,
Figure 2).
5
On this point, it should be noted that the theoretical motivation for privatization is generally
not explicitly mentioned in privatization reports. Hodges (1997) shows that the promotion of
competition is rarely stated as an objective of a privatization, and where it is, it can lead to
difficulties in drawing objective audit conclusions (Hodges, 1997). Prager (1994) suggests
that contracting out of government services to private actors “will neither reduce government
outlays nor increase government efficiency unless the decision makes economic sense. For
example, a government authority may decide that downsizing is a political, not financial or
economic, imperative, or outsourcing may serve as a threat to weaken the power of an
entrenched bureaucracy or labor union” (Prager, 1994).
7
competitive market to emerge (Levin & Tadelis, 2010)6. With a specific focus on healthcare,
hospital private ownership is not necessarily associated with a higher efficiency than public
ownership (Tiemann, Schreyögg, & Busse, 2012)7. Building on that premise, we argue that
privatization needs some additional conditions to enable its potential benefits. Enthoven
(2004) argued that market forces should be made strong enough to deliver efficient healthcare
systems (Enthoven, 2004), suggesting that privatization is more beneficial when the share of
the outsourced market is substantial. We hypothesize that privatization negatively moderates
the inverted U-shaped relationship between outsourcing and expenditure. In the case of INHS,
Regions might be attracted by privatization only if this decision is coupled with a substantial
level of outsourcing. In other words, privatization at low levels of outsourcing is not
beneficial, consistent with Enthoven (2004), but it leads to relatively higher savings when
used as a complement of outsourcing.
Hypothesis 2. Privatization negatively moderates the impact of outsourcing on public
expenditure (negative interaction term).
------------------------------------Insert figure 1 about here
------------------------------------The unattainable efficiency puzzle in the context of NHSs
According to our theoretical model, we would expect Regions to adopt extreme
configurations rather than intermediate ones. However, empirical data (cfr. Figure 3) show
strong evidence in support of mixed outsourcing and privatization levels, with smooth
6
Factors impeding competition are, for instance, problems in monitoring performance or the
need for flexibility.
7
This evidence is consistent with findings in other domains of public services, like
transportation (Zullo, 2008).
8
transitions over time. Therefore, the question arises of why Regions choose and maintain
(theoretically) inefficient models.
Two possible explanations can be envisaged. 1) Regions incorrectly choose their mix between
outsourcing and privatization or 2) reasons other than efficiency lead the choice of the mix
and determine the speed of adjustment. The first explanation relies on the assumption that the
appropriate model is either a fully public and in-house service or, alternatively, a fully private
and outsourced one. Neither of the two possibilities is applicable to the INHS. Accepting the
former case would imply rejecting NPM totally and excluding important stakeholders of the
NHS (e.g., private hospitals). Accepting the latter case would contradict the foundations of a
universalistic health system, requiring a dramatic institutional and political change. We
therefore believe that the second explanation is appropriate and that efficiency, even if
theoretically affordable, is an illusion in practice. To support this thesis we build on two
arguments. First, given that the initial choice is an intermediate one, we show that the
adjustments suggested by hypotheses 1 and 2 are smooth, and second, as a consequence,
efficiency-improvement strategies can be time-inconsistent or ambiguous.
The choice of the initial configuration
NHSs are characterized by the direct provision of services. However, they may decide to
outsource the management of some production processes to other providers (public, private or
public-private mixed entities), thus generating incentives to specialization in each outsourced
phase. In recent years, the necessity of reducing production costs has pushed universal
healthcare systems to reconsider the possibility of moving organizational boundaries by
exploiting specialization economies and inducing competition (by privatization) under both a
rationality8 constraint and a quality constraint9. Burke and Goddard (1990), building on
8
i.e., the savings from outsourcing are greater than the new transaction costs.
9
Transaction Costs Economics, justified the appropriateness of an intermediate level of
outsourcing. In particular, they verified the occurrence of five conditions10 that justify the
failure of a pure-market configuration of transactions in the British NHS. As these conditions
were highly dependent upon the peculiarity of the sector rather than of the country, their
conclusions can be fairly generalized to the Italian case11.
The first condition is bounded rationality, an issue that becomes increasingly important with
the growth of complexity in activities and technological advancements in healthcare. The
second one is opportunistic behavior: even if it “has not traditionally been dominant in
healthcare [as] there has been an emphasis on ethical behavior and moral commitment”
(Burke & Goddard, 1990), most actors involved in the resource allocation process had
reported experiencing opportunistic behaviors. Third, transactions are restricted in frequency
because of the limited number of contracting parties. This characteristic is highly desirable in
healthcare as strong standardization requirements are necessary to ensure equity within the
system. However, this comes at the costs of facing a greater bargaining power and
discretionality in the final choice12. Fourth, complexity is inherent to medical activity and, by
extension, to the whole NHS. This requires a leading actor who is accountable for
performance of the whole system. Finally, healthcare is also characterized by a significant
9
In Italy, minimum granted services (LEA, Livelli Essenziali di Assistenza) are defined and
listed.
10
Based on Williamson (1973, 1975, 1979, 1985), they refer to the conditions under which the
market is likely to be a less efficient allocator of resources. The first two are behavioral
conditions; bounded rationality and opportunism. The remaining three relate to (i) a low
frequency of transactions; (ii) high complexity and uncertainty surrounding transactions and
(iii) a high degree of enduring idiosyncratic investments.
11
Note that Rehnberg’s (1990) study of the organization of the Swedish health care system
concludes that the degree of asset specificity and of uncertainty are poor predictors of the
degree of externalization. Based on Transaction Costs Economics, he initially predicted that
the production of asset-specific and uncertain service would be internalized, while the
production of standardized services would be externalized. In fact, an integration pattern was
found in which not only asset-specific transactions were internalized, but so were the
transactions of standardized services offered by a large number of sellers.
12
This is also in light of uncertainty about demographic, epidemiological and technological
patterns and outcomes of care.
10
presence of idiosyncratic investments (staff, education, equipment) that are made in a quasi
monopsonistic market, requiring ad-hoc governance systems that are different from those of a
pure market (Williamson, 1979). All of these elements lead to the prediction that market is
not the most efficient organizational form for Health Systems. We therefore expect
intermediate levels of outsourcing and privatization to be the initial configuration, which may
differ across Regions depending upon the strength and the manner in which these conditions
appear and interact together.
Following hypotheses 1 and 2, at intermediate levels of outsourcing and privatization, the
opportunity exists to improve efficiency through mix adjustments. However, adjustments in
complex systems are not a straightforward process.
The smoothness of change
So far, we have relied substantially on Transaction Costs Economics in explaining why
Regions choose inefficient initial organizational configurations. To explain the smooth
dynamic of change, we also consider a sociological argument. Transaction Costs Economics
is often criticized by organizational sociologists because it operates with an under-socialized
conception of human actions (Ferlie, 2007). Thus, the governance structure and contracting
rather than social relationships are the driving forces for decisions. In other words, new
institutional economics ignores the proposition that behaviors are embedded in social
relationships (Granovetter, 1985). Embedded relationships make social actors interdependent;
therefore, any status quo shows more rigidities than do the ones we expect in an undersocialized world. Adding the social perspective to our analysis allows us to explain the
gradual evolutionary pattern of make-or-buy and privatization in Regional Health Services
(RHSs). At any point in time, the potentially efficient configuration is not immediately
11
reachable because it would require a dramatic change in the institutional and relational
framework. Therefore, small changes rather than dramatic ones are expected in RHSs. This is
also consistent with Lindblom’s incrementalist theory of decisions in public management.
Ambiguity and time-inconsistency
According to hypotheses 1 and 2, it is likely that given the initial condition, a dominant
strategy may not exist. This happens, for example, when a Region is near the maximum point
of the outsourcing-expenditure curve. Under such a condition, either an increase or a decrease
in outsourcing may induce efficiency. Moreover, in the real world, the final impact of a policy
is subject to uncertainty. Finally, even if a more convenient strategy is identifiable, it may
carry costs other than the current production-related ones, such as the costs associated with
changing the institutional framework. For example, insourcing back some activities
determines the necessity of remodeling the supply structure by building new public hospitals
and establishing relationships with additional suppliers. The above-mentioned points suggest
that efficiency-improving strategies are potentially ambiguous.
In addition, smooth strategies may be time-inconsistent, which means that the objective
configuration changes with the process of approaching it. A smooth process generates a
repositioning of the initial condition at every point in time. As a consequence, new efficiencyimproving strategies may emerge that are not necessarily consistent with the process followed
up to that point. Time-inconsistency may also derive from institutional and political factors.
The former are, for example, the costs of downsizing the level of fixed production factors or
of building a new set of agreements with other counterparties. Political factors, instead, are
related to the need to increase the consensus in specific phases of political cycles.
12
From the above, we can conclude that i) both privatizing and modifying outsourcing levels
can be considered to be available policies for improving efficiency, ii) as mechanisms
explaining improvements are different, these two choices can be considered separate, iii)
Regions may start from different intermediate configurations, iv) the modification of
decisions about privatization and outsourcing levels are embedded in preexisting social
relationships/structures, making change characterized by inertia and v) efficiency improving
strategies may be ambiguous and time-inconsistent.
Therefore:
Proposition 1. Regions choose intermediate levels of outsourcing and privatization, but with
different mixes.
Proposition 2. Regional strategies may change over time as predicted by hypotheses 1 and 2,
but they rarely experience dramatic changes that would require a strong revision of the
preexistent relationships between social public and private actors.
Proposition 3. Regional strategies may not follow consistent patterns over time due to the
ambiguity and time-inconsistency of optimal strategies.
As predicted by proposition 1, we expect to find a distribution of choices that could be
classified and interpreted on the basis of the prevalent presence of attributes, according to
Figure 2:
13
------------------------------------Insert figure 2 about here
------------------------------------METHODS
We first proceed by testing hypotheses 1 and 2, and then we compare the predictions from our
simulations with the real evolution of outsourcing and privatization in some of the Regions,
interpreting apparent inconsistencies in the light of propositions 1 through 3. We combine,
therefore, methods of quantitative statistical analysis with qualitative interpretive analysis.
Statistical analysis
The Italian NHS represents an excellent setting for studying outsourcing and privatization. In
2001, the Italian government approved legislation13 that recognized the 21 Italian Regional
Health Services as accountable for healthcare expenditure on an annual budget allocation and
gave them substantial autonomy in the definition of policies for attaining a sustainable level
of efficiency14. At the beginning of the decentralization/regionalization process, Italian
Regions showed very different starting points in terms of the urgency of turnaround and the
solutions already in place. Therefore, there is substantial variation both over time and across
Regions (cfr Figure 3).
13
L. 405/2001.
For example, in accordance to national thresholds, RHSs can independently increase /
decrease the level of resources devoted to healthcare varying the level of regional taxes and
co-payments for services directly paid by service users to regional governments.
14
14
To test our hypotheses, we use the public account data of the Local Health Authorities
(LHAs) in the 21 Italian Regions over a period of 8 years (2002-2009)15. Our dependent
variable is the natural logarithm of the per capita public healthcare expenditure16. Health
expenditure is measured as the total regional health expenditure for each year. Outsourcing is
measured as the proportion of hospital beds that are non-directly managed by LHAs (i.e.,
managed by public hospital enterprises cfr. Lo Scalzo, Donatini, Orzella, Profili & Maresso,
2009) and privatization is measured as the proportion of beds that are owned by private
hospitals for which a contract has been signed with the Region or the LHA (according to the
institutional setting). Controls include the level of regional healthcare deficit for every year,
the presence of a healthcare turnaround plan and the local government party affiliation,
together with the regional fixed effects and specific dummy variables for 200517. We identify
the effect with a linear panel regression model with regional fixed effects.
The variable definitions and sources are summarized in Table 1.
------------------------------------Insert table 1 about here
------------------------------------In the second part our empirical analysis, we interpret the evolution of regional choices
considering the interplay of two explanatory forces: the first one is the opportunity for
improving efficiency (simulation exercise in our statistical analysis), and the second one is
represented by the consequences of the smoothness of change (cfr. supra), such as ambiguity
and time-inconsistency.
15
We excluded year 2001 because the decentralization law was approved in November.
We used the logarithmic transformation for interpreting model coefficients as percent
variations.
17
In 2005, an important innovation in the categorization of hospital bed ownership occurred.
In particular, hospital beds of public research-hospitals (IRCCS) were counted together with
public hospital beds, while they were previously excluded.
16
15
RESULTS AND DISCUSSION
As noted above, the case of the Italian RHSs shows sufficient variability with respect to the
attributes depicted in Figure 2; thus, we expect to find at least 1 Region per sector in the
matrix. In particular, we were able to position the 21 RHSs in the matrix based on two main
dimensions: the proportion of hospital beds managed by private entities and reimbursed by
the RHS (model of privatization) and the proportion of hospital beds managed by public and
private entities other than the Local Health Authorities (model of outsourcing). Figure 3
shows the positioning of the Regions in an outsourcing/privatization matrix in the observed
period, following the dimensions presented in Figure 2.
------------------------------------Insert figure 3 about here
-------------------------------------
According to our classification, Regions appear to be predominantly public service-based
with low levels of outsourcing (model 1 in Figure 2), but just one Region (Valle d’Aosta)
shows a complete direct provision of services by the LHA and, as a consequence, no presence
of private providers. This is consistent with our theoretical argument about the choice of
intermediate rather than extreme configurations. Lazio and Campania present a rather high
level of privatization, even with substantial variability across time (model 4 in Figure 2), as
do Lombardia and Calabria (not in Figure 3). Two Regions are consistent with model 2 in
Figure 2, i.e., outsourcing within the public sector.
16
Statistical analysis
Table 2 presents the correlations among the variables.
------------------------------------Insert table 2 about here
------------------------------------We estimated four models using linear panel data analysis (Region/year) with regional fixed
effects (Table 3). We centered our explanatory variables over the regional means across the
observed time period to reduce multicollinearity issues, and we estimated robust standard
errors to account for possible heteroscedasticity. In model 1, only control variables are
included. They explain 28.7% of the within-group variability. It is interesting to note that
Regions with turnaround plans show a higher level of health expenditure. Left-affiliated
governments tend to generate a 0.7% higher per capita expenditure, but the effect is not
significant in models 3 and 4. In model 2, we added the 2005 dummy variable, which is not
significant. Our explanatory variables are included in models 3 and 4. In model 3, we only
look at the direct effects, while the outsourcing squared and the interaction terms are added in
model 4. In model 4, we also excluded the 2005 dummy variable, which was not significant in
model 3. As the R-squared is substantially higher in model 4 (35.3%), we use this model to
test our hypotheses.
The linear coefficient for outsourcing is positive and significant, while the squared term’s
coefficient is negative and significant. Within our sample, we found evidence for both the
increasing and decreasing part of the relationship, supporting hypothesis 1. Hypothesis 2
predicted a negative interaction term. In model 4, the coefficient is negative and significant,
consistent with hypothesis 2.
17
------------------------------------Insert table 3 about here
------------------------------------Privatization shows also a significant and positive direct effect on expenditures. At small
levels of outsourcing, privatization is therefore likely to increase transaction costs more than it
is to generate savings. However, the negative impact of privatization on efficiency is then
gradually reduced with higher propensities to outsource. The overall effect on public
expenditure turns out to be negative after a certain level.
Interpretation of the actual dynamic
We now compare the suggestions based on simulations from model 4 with the actual dynamic
in three Regions: Piemonte, Sicilia and Lazio, which are instances of the models of public
production, public outsourcing and privatization, respectively (cfr. Figure 2).
Piemonte Region
From the starting point in 2002, there were different strategies for reducing public
expenditure. According to the simulation in Figure 4.a, the most effective ones suggest to
increase both outsourcing and privatization (long-dashed line; +50% privatization) or to have
a dramatic reduction in the presence of private actors coupled with a reduction in the
outsourcing level (dotted line, 0% privatization). In this case, ambiguity is evident, and no
clear-cut suggestion emerges from the simulation. Looking at the actual dynamic, the regional
government has increased both the levels of privatization and outsourcing from 2002 to 2003,
but the opposite strategy has been followed from 2003 to 2005. In the 2006-2009 period, the
Region increased both outsourcing and privatization again. Taken individually, both strategies
were consistent with the indications from the simulation exercise. However, we find evidence
of smooth change, with rather small yearly variations in the mix. As we discussed in the
18
theory section, time-inconsistency is one of the consequences of smoothness of change.
Moreover, time-inconsistency may also derive from institutional and political factors. In this
case, the right government was replaced in April 2005 by the left-wing coalition. The sharp
reduction of privatization in 2004-2005 may be interpreted as a choice to build an electoral
consensus by expanding the boundaries of public service18. The new left-party government
has resumed the outsourcing policy without giving excessive space to privatization. It is likely
that after observing more political cycles, the pattern of evolution could show recursive
dynamics. If this is true, then the potential levels of efficiency might never be reached because
the smoothness of change, the institutional rigidities and the length of political cycles do not
favor a long-term strategy.
Sicilia Region
In 2002, Sicilia Region showed higher levels of outsourcing as compared to Piemonte but
lower levels of privatization. The suggestions coming from the simulation exercise, even if
still subject to ambiguity, show a relatively sharper reduction in public expenditure when
privatization and outsourcing are both increased (long dashed line; +50% privatization) with
respect to decreasing levels of outsourcing and the exclusion of private providers (dotted line,
0% privatization). Coherent with the suggested strategy, the Region increased the level of
privatization by 14.5% from 2002 to 2003, creating the conditions for a greater effectiveness
of the outsourcing policy, and from 2006 to 2009, it favored the increase of outsourcing via
privatization. During these four years, the Region followed a balanced increase in its
outsourcing and privatization levels even though the following sharp changes occurred: the
adoption of a turnaround plan because of a high Regional healthcare deficit (2007) and the
election of a new government with the appointment of the right-wing party (2008). This
18
Note that this choice is traditionally considered proper for the left coalition, and the fact
that it was the right party that made it can be a signal of consensus-building behavior.
19
evolution pattern reaffirms the validity of the smoothness of change in complex and highly
institutionalized settings.
Lazio Region
In Regions such as Lazio, which shows a high level of privatization and a high-intermediate
level of outsourcing, reducing the presence of private hospitals could be detrimental in terms
of the productive efficiency according to hypothesis 2, which would suggest a further increase
in outsourcing. Looking at the simulation, from the starting point in 2002 (the intersection of
the curves), a better level of efficiency could be achieved by either increasing or decreasing
outsourcing. In the former case, an increase in privatization is also suggested. Looking at the
policy adopted by the regional government, it seems that during the first four years (20022004), the strategy suggested by the simulation was chosen, but then a reduction in the mix of
outsourcing and privatization was set during the government election campaign (2005-2006).
Insourcing back some activities again suggests a trigging intervention to promote the preelectoral consensus. The new left-wing party revamped a sharp expansion toward
privatization (2007) and an increase in outsourcing (2008-2009). The dynamic process of the
reconfiguration of the outsourcing and privatization levels generally operate as two separate
choices, with regional governments experiencing small but continuous modifications in the
levels of outsourcing.
------------------------------------Insert Figure 4 about here
------------------------------------We can conclude that Regions have shown that efficiency-improving strategies are identified
and smoothly implemented but are changed before their completion, generating non-linear
patterns of evolution. A general interpretation of these patterns is to be made in the light of
20
the fundamental question about the existence of steady-states. Are there (sooner or later)
reachable efficient configurations? Or, instead, is this instability moving between waves of
insourcing and outsourcing perpetually, with no long-lasting matching between problems and
solutions?
Arguments in support of the first interpretation may be built upon incrementalism and by
using the concept of punctuated equilibria19. Adopting this theoretical lens, efficiency is in
principle obtainable in the long run, but the path is not short or linear, and the excessive
amount of time may be perceived as a signal of impossibility. It is still possible that in some
cases a Region is not able to make decisions that would require a drastic change in the model,
for example because of institutional and social rigidities. This is the case, for example, in
Sicilia Region. Instead, if punctuated equilibria exist, it is possible that time-inconsistency
emerges to a greater extent. This is the case, for example, in Lombardia Region, where the
privatization-only process occurred between 1997 and 2001 and was coupled in the 20012003 period with a dramatic increase in outsourcing. In the following years, and notably since
2005, the strategy has been changed every year, without showing an identifiable pattern. In
fact, in 2005, some of the activities outsourced to other public institutions were insourced
back, while the remainder were privatized. In 2006, both outsourcing and privatization were
reduced. In 2007, there was a substantial downsizing of privatization, and in 2008-2009,
outsourcing and privatization were again increased. A useful framework for interpreting this
evolution is the garbage can model. Indeed, at least two of the typical conditions that favor the
emergence of garbage cans are met: in our case, both the technology is unclear (there is no
clear rule to reach efficiency, if we reject the incrementalist view), and preferences are
problematic to the extent that beyond the objective of efficiency, quality constraints and other
19
Punctuated equilibria in social theory is a method of understanding change in complex
social systems, particularly how policy change and the development of conflicts seem to
progress in extended periods of stasis, punctuated by sudden shifts with radical change.
21
relevant dimensions of performance do not allow for the identification of a univariate
objective function. Fluid participation, the third basic property of systems where garbage can
emerge, is not very applicable at such a macro context, but it is still observable, extending the
analysis to the whole network of stakeholders instead of focusing only on the appointed
policymaker. According to this view, no efficiency is obtainable by definition of the dynamics
because regional strategies for improving efficiency will be inherently chaotic, highly
unpredictable and ineffective in the long run.
CONCLUDING REMARKS
Adopting different theoretical interpretive approaches, we have shown that efficiency,
intended as a final steady-state, is an illusion. Some drawbacks should, however, be noted.
The choice of disentangling outsourcing and privatization is theoretically intriguing but quite
new, and there was little possibility of identifying an empirical setting with available data
where the two choices were not coincident. Consequently, we made some simplifying
assumptions in the definition of our variables. In the case of the INHS, public outsourcing is
not conceptually homogeneous across Regions. In particular, quasi-market systems differ
considerably in the degree of separation (institutional and financial) between hospitals
directly managed by LHAs and public hospital enterprises. Therefore, our measure is a
potentially biased proxy of public outsourcing and should be taken with care when discussing
normative implications. However, the focus of this paper is not the identification of an
optimal policy, but rather the interpretation of public choices that deviate from a theoretically
efficiency-improving consistent strategy. Another possible concern is the generalizability of
the results. The health sector shows significant peculiarities with respect to both the content
and the country. However it can be considered a relevant case for i) the possibility of
disentangling privatization and outsourcing and ii) for the substantial weight of public
healthcare expenditure on total government expenditure (15% on average in European
22
countries and 21% in the U.S. in 201020). Our study sheds new light on the gap between the
normative implications derived from comparative statics exercises, based on statistical
analyses, and the actual, apparently unsystematic, dynamic of public policies.
References
Anessi-Pessina, E., Cantù, E., & Jommi, C. 2004. Phasing out market mechanisms in the
italian national health service. Public Money & Management, 24(5): 309-316.
Anessi-Pessina, E., & Steccolini, I. 2005. Evolutions and limits of new public Management—
Inspired budgeting practices in italian local governments. Public Budgeting & Finance,
25(2): 1-14.
Armeni, P. 2011. La spesa sanitaria: Composizione ed evoluzione (Healthcare expenditure:
Composition and evolution). In E. Cantù (Ed.), Rapporto OASI 2011: L'aziendalizzazione
della sanità in italia: 1-70. Milano: Egea.
Ashenfelier, O., & Card, D. 1985. Using the longitudinal structure of earnings to estimate the
effect of training programs. Review of Economics & Statistics, 67(4): 648.
Battaglio Jr., R. P., & Ledvinka, C. B. 2009. Privatizing human resources in the public sector.
Review of Public Personnel Administration, 29(3): 293-307.
Bloching, B., Stock, H. F., & Scheel, J. 2008. Hospital marketing should focus on physicians:
Lessons from germany. Journal of Medical Marketing, 8(4): 303-310.
Borgonovi, E., Fattore, G., & Longo, F. 2012. Management delle amministrazioni pubbliche
(2nd ed.). Milano: Egea.
20
Source: OASI Report, 2012.
23
Boyne, G. A. 1998. Bureaucratic theory meets reality: Public choice and service contracting
in U.S. local government. (cover story). Public administration review, 58(6): 474-484.
Brekke, K. R., Holmas, T. H., & Straume, O. R. 2011. Reference pricing, competition, and
pharmaceutical expenditures: Theory and evidence from a natural experiment. Journal of
Public Economics, 95(7): 624-638.
Breunig, C., & Koski, C. 2012. The tortoise or the hare? incrementalism, punctuations, and
their consequences. Policy Studies Journal, 40(1): 45-68.
Buchanan, R. J., Cappelleri, J. C., & Ohsfeklt, R. L. 1991. The social environment and
medicaid expenditures: Factors influencing the level of state medicaid spending. Public
administration review, 51(1): 67-73.
Burke, C., & Goddard, A. 1990. Internal markets - the road to inefficiency? Public
Administration, 68(3): 389-396.
Chalkley, M., & Malcomson, J. 1998. Contracting for health services with unmonitored
quality. Economic Journal, 108(449): 1093-1110.
Cohen, M. D., March, J. G., & Olsen, J. P. 1972. A garbage can model of organizational
choice. Administrative Science Quarterly, 17(1): 1-25.
David, G., & Chiang, A. J. 2009. The determinants of public versus private provision of
emergency medical services. International Journal of Industrial Organization, 27(2): 312319.
Enthoven, A. C. 2004. Market forces and efficient health care systems. Health affairs, 23(2):
25-27.
24
Ferlie, E. 2007. The creation and evolution of quasi markets in the public sector: A problem
for strategic management. Strategic Management Journal, 13(S2): 79-97.
Ferrario, C., & Zanardi, A. 2011. Fiscal decentralization in the italian NHS: What happens to
interregional redistribution? Health Policy, 100(1): 71-80.
Ferrè, F., Cuccurullo, C., & Lega, F. 2012. The challenge and the future of health care
turnaround plans: Evidence from the italian experience. Health Policy, 106(1): 3-9.
Granovetter, M. 1985. Economic action and social structure: The problem of embeddedness.
American journal of sociology: 481-510.
Guerrero, I., Mossé, P. R., & Rogers, V. 2009. Hospital investment policy in france: Pathways
to efficiency and the efficiency of the pathways. Health Policy, 93(1): 35-40.
Hefetz, A., & Warner, M. E. 2012. Contracting or public delivery? the importance of service,
market, and management characteristics. Journal of Public Administration Research &
Theory, 22(2): 289-317.
Hodges, R. 1997. Competition and efficiency after privatization: The role of the NAO. Public
Money & Management, 17(1): 35-42.
Holzer, M., & Yang, K. 2004. Performance measurement and improvement: An assessment of
the state of the art. International Review of Administrative Sciences, 70(1): 15-31.
Kettl, D. F. 2006. Managing boundaries in american administration: The collaboration
imperative. Public administration review, 66: 10-19.
Levin, J., & Tadelis, S. 2010. Contracting for government services: Theory and evidence from
U.S. cities. Journal of Industrial Economics, 58(3): 507-541.
25
Lindblom, C. E. 1959. The science of "muddling through". , 19(2): 79-88.
Lindblom, C. E. 1979. Still muddling, not yet through. Public administration review, 39(6):
517.
Lynn Jr., L. E. 1998. The new public management: How to transform a theme into a legacy.
Public administration review, 58(3): 231-237.
Mailly, R. 1986. The impact of contracting out in the NHS. Employee Relations, 8(1): 10-16.
Mapelli, V., & Lucioni, C. 2003. Spending on pharmaceuticals in italy: Macro constraints
with local autonomy. Value in health : the journal of the International Society for
Pharmacoeconomics and Outcomes Research, 6 Suppl 1: S31-45.
Murray, J. G., Erridge, A., & Rimmer, N. 2012. International lessons on austerity strategy.
International Journal of Public Sector Management, 25(4): 248-259.
Newcomer, K. E. 2007. Measuring government performance. International Journal of
Public Administration, 30(3): 307-329.
Nicholson Crotty, S., Theobald, N. A., & Nicholson Crotty, J. 2006. Disparate measures:
Public managers and Performance‐Measurement strategies. Public administration review,
66(1): 101-113.
Pendleton, A. 1999. Ownership or competition? an evaluation of the effects of privatization
on industrial relations.. Public Administration, 77(4): 769-791.
Pollitt, C. 2006. Performance management in practice: A comparative study of executive
agencies. Journal of Public Administration Research and Theory, 16(1): 25-44.
26
Prager, J. 1994. Contracting out government services: Lessons from the private sector. Public
administration review, 54(2): 176.
Riccucci, N. M. 2001. The 'old' public management versus the 'new' public management:
Where does public administration fit in? Public administration review, 61(2): 172-175.
Scalzo, A. L., Donatini, A., Orzella, L., Profili, A., & Maresso, A. 2009. Italy. health system
review. Health Systems in Transition, 11(6).
Tiemann, O., Schreyögg, J., & Busse, R. 2012. Hospital ownership and efficiency: A review
of studies with particular focus on germany. Health Policy, 104(2): 163-171.
Vernon, J. A. 2002. Drug research and price controls. Regulation, 25(4): 22.
Williamson, O. E. 1973. Markets and hierarchies: Some elementary considerations. American
Economic Review, 63(2): 316-325.
Williamson, O. E. 1979. Transaction-cost economics: The governance of contractual
relations. Journal of Law & Economics, 22(2): 233-261.
Zullo, R. 2008. Transit contracting reexamined: Determinants of cost efficiency and resource
allocation. Journal of Public Administration Research & Theory, 18(3): 495-515.
27
Tables and Figures
Figure 1: Theoretical model
OUTSOURCING
-U
PUBLIC
EXPENDITURE
PRIVATIZATION
Figure 2: Models of privatizing and outsourcing
Public ownership
Private ownership
Model 1: Public Production
Model 3: n.a. in NHS
Make
Model 2: Public Outsourcing
Buy
(separation
of
production
Model 4: Privatization
activities and service design
within the public sector)
28
Figure 3: 2002-2009 range of Regional choices of outsourcing and privatization
Private ownership
0
Make
Make
Public ownership
Trento
Molise
Veneto
Sardegna
Marche
Toscana
Basilicata
Umbria
0.5
Puglia
1
Piemonte
1.5
0
0.2
0.4
Friuli VG
0.6
0.8
1
2
Campania
2.5
Lazio
1.2
Sicilia
Buy
Buy
3
3.5
Public ownership
Private ownership
Notes:
- The intersection of axes is arbitrary and has only descriptive value;
- Yearly observations have been connected with a smooth line to facilitate the
identification of Regions;
- In this Figure, we do not report years. Even if we lose the direction of change, we are
interested in mapping the overall range of Regional variation of the outsourcing and
privatization mix. The dynamic perspective will be analyzed in the results and
discussion section.
- Some Regions (e.g., Lombardia) are not reported for scale reasons.
29
Figure 4: Actual dynamic of outsourcing/privatization mix (left graph) and simulated
response of public expenditure as a function of outsourcing at different levels of
privatization.
4.a: Piemonte Region
Actual dynamic
Simulation
Piemonte Region
1.20
Outsourcing
1.24
2005
1.26
2004
2006
1.28
2003
1.30
2002
2007
2008
1.32
1.34
2009
1.36
1.38
0.30
0.32
0.34
0.36
0.38
0.40
0.42
Piemonte Region
4500
Per-capita public health expenditure
1.22
0.44
0.46
-60
4000
3500
3000
2500
2000
1500
1000
500
0
-50
-40
-30
-20
-10
0
10
20
30
40
50
60
50
60
50
60
outsourcing (centered)
Privatization
Average priv.
-50%
+50%
0
4.b: Sicilia Region
Actual dynamic
2.6
Sicilia Region
2003
2004
2.8
2006
2005
2007
2.9
2008
3
2009
3.1
3.2
0.25
0.27
0.29
0.31
0.33
0.35
0.37
0.39
0.41
Sicilia Region
3000
Per-capita public health expenditure
2002
2.7
Outsourcing
Simulation
0.43
-60
2500
2000
1500
1000
500
0
-50
-40
-30
-20
Privatization
-10
0
10
20
30
40
outsourcing (centered)
Average priv.
-50%
+50%
0
4.c: Lazio Region
Actual dynamic
2.0
Simulation
Lazio Region
Lazio Region
5000
Per-capita public health expenditure
2.1
2.2
2007
Outsourcing
2.3
2.4
2002
2008
2.5
2003
2.6
2006
2.7
2.8
2009
2005
2004
2.9
3.0
0.60
0.70
0.80
0.90
Privatization
1.00
1.10
-60
4500
4000
3500
3000
2500
2000
1500
1000
500
0
-50
-40
-30
-20
-10
0
10
20
30
40
outsourcing (centered)
Average priv.
-50%
+50%
0
30
Note: Outsourcing has different scales in the two graphs (the simulation is based on centered
outsourcing). The intersection of the curves in Figure 4.b represents the starting point in
2002. The four curves represent different levels of privatization, starting from the Regional
average, centered at 2002. The vertical axis reports values in the inverse order, for
consistency with Figure 3.
31
Table 1: Variable definitions and sources
Measure
Source
ln of per capita public
healthcare expenditure
Reports of the
Osservatorio
sull’Aziendalizzazione
della Sanità in Italia
(OASI) (various
years)
VARIABLES
Per capita healthcare
expenditure (ln)
Outsourcing
Privatization
Turnaround plan
Left
Proportion of hospital beds
non-directly managed by
LHAs
Proportion of hospital beds
managed by private hospitals
1 if Region i has an active
turnaround plan at time j
Number of months in the year
with a left-affiliated Regional
government
Deficit
Per capita health deficit (€)
year 2005
1 if year = 2005
Observations
Number of Regions
168
21
Ministry of Health and
OASI observatory
Ministry of Health and
OASI observatory
OASI reports (various
years)
Ministry of internal
affairs
OASI reports (various
years)
Table 2: Correlation Matrix
Variables
(1)
(1)
Per capita healthcare expenditure (ln)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
1
(2)
Outsourcing (centered)
0.15
1
(3)
Outsourcing (centered)^2
-0.07
0.01
1
(4)
Privatization (centered)
-0.03
0.02
-0.01
1
(5)
Private (centered) x Outsourcing (centered)
-0.12
-0.22
0.62
-0.05
1
(6)
Turnaround plan
0.31
0.00
-0.06
0.03
-0.04
1
(7)
Left
0.14
0.00
-0.24
0.12
-0.16
-0.06
1
(8)
Deficit
0.26
-0.03
-0.07
0.06
-0.07
0.29
0.07
1
(9)
year 2005
0.19
0.08
0.03
-0.01
0.05
-0.09
0.07
0.13
32
Table 3: Estimation results
Per capita public health expenditure (ln)
(1)
(2)
(3)
(4)
0.0013**
(0.0003)
0.0092**
(0.0011)
-0.0002**
(0.0000)
0.3123*
(0.1511)
-0.05919**
(0.0053)
0.1262**
(0.0206)
0.0055
(0.0035)
0.0001**
(0.0000)
VARIABLES
Outsourcing (centered)
Outsourcing (centered)^2
0.2691†
(0.1523)
Privatization (centered)
Private (centered) x Outsourcing (centered)
Turnaround plan
Left
Deficit
year 2005
Constant
Observations
R-squared
Number of Regions
0.1451**
(0.0187)
0.0068*
(0.0031)
0.0001**
(0.0000)
0.1481**
(0.0191)
0.0066*
(0.0032)
0.0001**
(0.0000)
0.0160
(0.0137)
7.32200** 7.3216**
(0.0210)
(0.0209)
0.1318**
(0.0206)
0.0056
(0.0034)
0.0001**
(0.0000)
0.0147
(0.0145)
7.3318**
(0.0226)
168
0.28
21
168
0.31
21
168
0.29
21
7.3344**
(0.0220)
168
0.35
21
Robust standard errors in parentheses
†
p < .10
* p < .05
**p < .01
33