Ceris-Cnr, W.P. N°3/2007

Taroni, 2005). The sample includes two types of hospitals: those directly managed by the LHU (ASL from
now on), and other major hospitals that have been
hived off from the LHU and transformed into independent enterprises called Aziende Ospedaliere (AO
from now on).
This unique dataset includes all the publicly owned
firms involved in the provision of hospital services in
the Piedmont Region. The time span covered by the
data follows the different reforms of the National
Health Service (NHS), so that our units are affected by
the policy of downsizing of the industry which has
been pursued during the 90s, and which is still
regarded as one of the primary areas of intervention to
control health expenditure. Planning at the regional
level of health care provisions (as envisaged in the
recent Piedmont Socio-Health Plan for the years 20062010) foresees a reorganisation of the regional hospital
network, with the aim of increasing the quality and the
effectiveness of services. This would imply a reduction
of the required number of beds, due to the planned
reduction of average length of stay, and a parallel
increase in outpatient treatments, home care services,
consultancy and day hospital treatments.
Information on the number of beds and on the quantity and complexity of the services provided (number
of patients, average DRG weight, number of inpatient
days) have been collected for each single hospital
within a LHU and for each AO. The total number of
beds, both ordinary and for day-hospital, are then computed for each ASL by aggregating the values of the
different hospitals which belong to the same LHU. Unfortunately, disaggregated information on the costs and
on the labour force are available only for AO, but are
not available for each hospital within each LHU. This
limitation can represent a problem for ASL units, since
staff costs can be related also to community services,
rather than hospital services. For such units, considering all costs as relative to the core hospital activity
would be inappropriate, so that caution must be put in
choosing which type of cost can be included in the
study. To that purpose, the different types of costs have
been selected and reorganised so as to obtain a measure
of operating cost with a composition that can be comparable for ASL and AO structures. First, financial
costs, extraordinary and atypical costs have been subtracted. The breakdown of the remaining costs is
shown in table 1. As can be easily seen, the cost structure is rather different between ASLs and AOs. If labour costs (in particular medical staff) represent 50%
of total operating costs for AO, in the case of ASL
their share is only 35% in 2000-2002 and 25% in the
last two years of observation. On the other hand, a
large portion of costs of ASL structures is relative to

outsourced services (more than 60% in 2004), a category that is not so important (less than 10% of costs)
for AOs. The share of the costs of drugs is about 3%
for ASLs and 6.5% (increasing up to 8.3% in 2004) for
AOs. The relative importance of operating services
given out by contract (such as food services, cleaning
and laundry) is different among the two types of hospitals too: it is about 2% for ASLs, and 4-5% for AOs.
Finally, depreciation and administrative expenditures
weight less for the former than for the latter.
The figures in table 1 clearly confirm that the two
types of hospitals are not performing identical tasks.
Since our aim is to identify an operating cost structure
which is as much homogeneous as possible, we selected the costs items that are more closely related to
the core activity of hospitals, that is the provision of
health care services. We come out with a final aggregation named operating hospital costs (OHC, the dependent variable in our econometric model) which is the
sum of the costs of the following inputs: labour, drugs,
capital (the measure of which is proxied by the total
number of beds)1. As shown in table 2, for what concerns the relative weight of the different cost categories, the two types of hospitals are now much more
similar. Labour costs are about 86% of operating hospital costs, while the weights of drugs and depreciation
are respectively 9.6% and 4.4%. OHC has an average
value of 79 million euro for ASLs (average yearly
growth rate of 3.6%) and 122 million euro for AOs
(average yearly grow rate of 4.8%).
2.2. Explanatory variables of the cost model
Exploiting the informative content of the database, we
have obtained the following explanatory variables to be
included in the estimation of the cost function: output,
complexity of provided services (case-mix), input prices. The full sample is a panel of 29 productive units
which are observed over a period of 5 years, for a total
of 145 observations. As an index of production volume
(Y) we opted for the total number of patients per year
(ordinary and in day-hospital). In addition, in order to
keep into account the severity of illnesses, a control
variable of the average DRG weight (DRGW) has been
added. Such a variable should reflect the differences in
the production mix, i.e. the average degree of complexity of the services provided by the hospital structures2.
1

Such a restricted cost aggregate corresponds on average to
32% of total operating costs for ASL (65% for AOs).
2
For example, a tonsillectomy is a typical operation with a
low degree of complexity (DRG weight 0.27), while thyroid and
cardiovascular operations have an average (DRG weight 1.04)
and a high degree (DRG weight 2.40) of complexity,
respectively.
7