Quality & Quantity 35: 33–48, 2001.
© 2001 Kluwer Academic Publishers. Printed in the Netherlands.
On Long-term Forecasting of Social Security:
A Robustness Analysis
TATIANA ERMOLIEVA, LANDIS MACKELLAR and ANDERS WESTLUND
International Institute for Applied Systems Analysis (IIASA), A-2361 Laxenburg, Austria
Abstract. In this paper we present a macro-economic demographic growth model having a special
focus on social security. It is designed to study the variability of responses of the system in presence
of risks and uncertainties. Here we analyze the robustness of the model towards uncertainties in
parameter speciﬁcations, introduced by ARCH-M models with the incorporation of intervention pro-
cesses. The parameters varied are labor force participation rates (one of the key sources of uncertainty
in the social security policy debate), and the parameters of the production function (the key source
of uncertainty in any long-run economic analysis). The sensitivity analysis focuses on two variables:
assets of the private pension system and the balance of the public ‘Pay As You Go’ pension system.
Special attention is given to convergence properties of the macro-economic model.
Key words: social security forecasting, structural changes, misspeciﬁcation, robustness
Uncertainties in forecasting the future of social security stem from a number of
sources, among them, uncertainty in demographic projections, household saving
rates and labor force participation rates. As in any long-term analysis, possible
changes in the production function, i.e., technological change, are of special con-
cern. With population aging now rapidly underway, the future of social security
regimes – the broad spectrum of institutions, public and private, which provide
for income in old age – is being questioned (OECD, 1998; World Bank, 1994).
Long-term projections and simulation analyses of pension system revenues and
expenditures, stocks of private pension savings, etc., play a large role in the policy
debate. In this paper, we report on the robustness analysis of a model designed for
We analyze the robustness of model solutions with respect to initial parameter
assumptions and consider the cases of parameter time invariance, i.e., robustness
of the model towards changing parameters. The parameters varied are labor force
participation rates (a key source of uncertainty in the social security policy debate)
and the parameters of the production function (the key source of uncertainty in any
Corresponding author: Tel: +46-87369231; Fax: +46-834161; E-mail: STAW@HSS.SE