II Jornadas de Análisis Input-Output. Crecimiento, Demanda y Recursos Naturales
Zaragoza del 5 al 7 de Septiembre de 2007
APPLICATION OF THE DELPHI TECHNIQUE IN THE PREPARATION OF THE 2001 INPUT-
OUTPUT TABLES FOR CATALONIA
Jordi Galter Pareja 1 , Javier Mariscal
Institut d’Estadística de Catalunya (Idescat), Via Laietana 58, 08005 Barcelona
This work presents the results of an innovative use of the Delphi technique in the preparation of the
Input-Output Tables for Catalonia (TIOC 2001). Using this technique for obtaining subjective
information from experts comes forward as a valid and reliable option for improving the quality of the
data with which the Input-Output tables are constructed. The application presented thus opens up an
interesting field of development for this technique and also gives several contributions for appraising
the execution of a Delphi exercise of this type. The goal of this application is to obtain the necessary
information for the construction of the intermediate consumption matrix that is not available through
other sources: the consumption of raw materials and external services as well as their geographical
source. The study was limited to the industrial sector.
Qualitative Delphi-type methodologies can be used to obtain the information necessary from an
external set of experts, in order to partially feed a quantitative economic model. This work reports on
the results of an experiment properly structured in this last direction: the use of the Delphi method to
obtain the information necessary for drawing up the input-output tables from experts. This application
was ground-breaking in Spain, no similar international experiments being known of.
The Delphi method is a social research technique which has the aim of obtaining a reliable group
opinion from a set of experts. This is a method of structuring communication between a group of
persons who can provide valuable aid for solving a complex problem. Its main characteristics are as
- This is an iterative process. The experts must be consulted at least twice on the same
question, so that they can rethink their reply, assisted by the information that they receive from
the opinions of the other experts.
- It keeps the anonymity of the participants, or at least that of their replies, as these go directly
to the coordinating group. This means one can undertake a process of group work with
experts who do not coincide in either time or space and also seeks to avoid the negative
influences that factors to do with the personality of the participating experts may have on the
- Controlled feedback. The exchange of information between experts is not free, but done
through the group coordinating the study, thus eliminating any information which is not
- Statistical response of the group. All the opinions form part of the final reply. The questions
are asked in such a way that a quantitative and statistical treatment of the replies can be
Input-Output tables (Leontief, 1966) are a statistical-accounting instrument representing the set of
economic relations taking place between the different agents in a territory in a particular period. Their
use is widespread both for statistical and analytical purposes.
For creating the intermediate consumption table one thus has to know the production functions (inputs
with their corresponding origins and proportions) of homogeneous products in the economy in
question, in the present case, in the economy of Catalonia.
* e-mails: Jordi Galter (email@example.com), Javier Mariscal (firstname.lastname@example.org)
This information has traditionally been obtained through postal surveys sent to random samples of
companies. In this project we opted mainly for using the Delphi method as an alternative to the
standard procedure of opinion polling and statistical inference for three main reasons:
- To improve the quality of the information provided by the companies. The traditional system
guarantees the number of replies, but not the quality of these. The Delphi method meant that
referral was made directly to the business experts in the best position to know such specific
information as that of the production functions of their products, as well as the origin of their
inputs and of the services used. These were normally the purchasing directors, assisted by
the financial directors.
- To eliminate wariness as regards the companies making known strategic and reserved
information such as the cost structures of their products. This was attempted by asking about
the characteristic production function of the product in general, not their own in particular. This
should also help to obtain more reliable information.
- To reduce costs. A traditional survey has very high costs. It involves getting information from
at least 20000 companies (for breaking down into 200 homogeneous products) while with the
Delphi it was possible to get this information with a much lower number of companies (roughly
four per product)
The Delphi method has thus been the main way of obtaining information for estimating the production
functions of most of the relevant products in the Catalan economy.
The participation results are highly positive. Of the companies asked, 425 replied, normally through
their purchasing managers, which presumably implies very high knowledge and quality of the replies.
In the second round 85% of the experts who took part in the first replied, which represents a really low
non-reply or dropout percentage, given the working characteristics of this group and the voluntary
nature of their participation.
The results obtained with the Delphi method have supplied a good deal of the Input Output Table
2001. More specifically, this technique has been used to obtain the purchases of raw materials and
their sources for over thirty branches of activity i (from a total number of 122 published branches). It
can as a general rule be affirmed that the data obtained has been extremely useful and that this is of a
quality equal to or over that of the data that would have been obtained by the traditional survey
method and it has doubtlessly involved a lower economic cost to obtain this.
As a general rule it can be asserted from the comparison that the data provided by the Delphi is as
valid as or more valid than the data provided by other sources (in spite of having a sample nine times
Without questioning the satisfactory level of the results reached, we must acknowledge that this study
does involve certain limitations, mainly stemming from the conditions in which it had to be undertaken,
apart from the intrinsic ones involved in subjective judgment as a source of information and planning.
We now mention the two most relevant of these:
• Pre-set number of rounds. The fact of having limited resources and places conditioned the
need to pre-set a maximum number of rounds. The limitation may well have prevented us from
reaching a greater degree of consensus in most of the items planned. Nevertheless, it is fair to
state that there are very few Delphi studies made for professional purposes which last over
• Number of experts. The ground-breaking nature and the ambition of the scope of the study
make it advisable to provide this, through caution, with dimensions easy to handle by the
coordinating team and by the institution in charge. Nevertheless if we could have had a
greater number of experts for each product, the result might possibly have been more precise.
Other limitations, a priori more evident, such as the budget ones (the experts did not receive any
remuneration), the difficulty identifying and ensuring the cooperation of genuine experts in the
functions of production and costs of services or the experts’ lack of knowledge about the methodology
of Input-Output Tables and the Delphi technique, were not considered to have any significant influence
on the results, as they were compensated by the organisational effort and the interest and enthusiasm
devoted to this study by both the coordinating team of this work and the cooperating experts
In the light of the results obtained and of the dynamics of the process observed, the main conclusions
that can be gleaned from this study are as follows:
1.- The Delphi Method and in general the techniques that are intended to pick up, improve and
process the subjective information most relevant for a particular economic or social problem are a
valuable complement for the application of techniques and construction of models mainly stemming
from objective data.
2.- This type of qualitative techniques, properly applied, may contribute to improving the efficiency of
the quantitative techniques, by allowing them access to a new type of information, of subjective
nature, which is relevant for understanding and modelling the phenomenon studied. Furthermore, it
may also improve the scientific thoroughness of the techniques based on objective information, by
forcing the researchers to make the value judgements that they are obliged to form during the
research process explicit, systematic and grounded.
3.- The particular characteristics of the Delphi Method, (anonymity of the participants, no need for
coincidence in time or place of these, controlled feedback and statistical reply of the group) make this
a technique with great possibilities for application when the training resource needed is in the hands of
business executives, given that this type of experts tend to have major agenda limitations, as well as a
certain logical reluctance stemming from the individual and organisational competitive environment in
which these persons lead their lives and perform their professional business.
4.- In the specific case put forward of the application of the Delphi Method to the preparation of the
symmetrical table of the Input-Output 2001 tables for Catalonia, and in view of the values obtained in
the selected quality gauges, we feel that the final results attained in this study have an acceptable
level of reliability and validity. In our opinion, this technique can be satisfactorily used in obtaining
information for supplying the I-O tables, attaining improvements both in terms of quality of responses
and in savings in social, economic and organisational costs.
5.-It would nevertheless seem necessary to perform scientific methodological studies comparing the
quality of the results obtained by means of the Delphi technique with the ones achieved through using
classic sampling techniques which would enable the conclusions drawn from this work to be positively