Working Papers, Konjunkturinstitutet - National Institute of Economic Research
The Financial Block in the Econometric Model KOSMOS
The model includes seven sectors (Central Bank,
central government, banks mortgage institutions, private and social
insurance, non-financial business and households, foreign) and nine asset
categories (certificates, bonds, bank deposits, loans, equity, net foreign
assets, notes and coin, insurance savings and claims on the National
Savings Scheme). Emphasis was laid on modelling demand for bonds and
certificates, which in turn affects the bond rate and the money stock.
Sector demand for equity is - with two exceptions - determined exogenously,
due to problems with modelling yield on equity and changes in market value.
Despite the relatively large number of assets the model is rather
simple, any refinements having been left to subsequent model versions. The
approach is eclectic. At the core of it lies the portfolio balance model,
but the portfolio in question is defined narrowly to include only net
foreign assets, certificates, bonds and (bank deposit) money. The remaining
assets are assumed to be acquired for other reasons than pure portfolio
investment and their purchase is assumed to be effected before any
portfolio decision is taken. Furthermore, the portfolio choice is assumed
to take place in two steps, the first decision referring to the choice
between foreign and domestic assets.
The model`s data base
consists of a flow-of-funds matrix (or its stock-value counterpart) derived
from the Financial Accounts published by Statistics Sweden. Annual
Financial Accounts time series for 1986-94 were distributed by half-years
(and in some instances quarters) using other sources, in particular the
data compiled by the Central Bank. Despite much effort, we were not always
able to reconcile time series coming from different sources.
quality of the data is in some respects rather poor, although a major
effort has been made to compile a good statistical data base for the
project. One of the problems, but definitely not the only one, is the
uncertainty regarding the extent to which assets are reported at market
value as opposed to nominal value.
The poor quality of the data
and the limited number of observations available affected the estimation
strategy. While statistical inference often is difficult in small samples
(in particual when only asymptotic distributions for the test variables are
known), data problems and apparent measurement errors made test results the
more dubious. In formulating equations, theory and considerations relating
to the desired simulation properties of the model were given precedence
over test results. Furthermore, the OLS estimator was employed, since
instrumental variable methods can result in large small-sample bias when
instruments are correlated with the error term or are only weakly
correlated with the endogenous explanatory variables (cf. Bound, Jaeger and
Baker (1995)). Standard computer printout is shown for all the equations as
a general information for the reader.
A rather uncommon approach
was employed in order to improve the reliability of the estimates.
Regressions were - whenever possible - based on quarterly data, the
resultant equations being subsequently transformed into semi-annual form to
conform with the requirements of KOSMOS. To this end, a theory of temporal
aggregation of equations was developed in Ruist (1996)).
structure of the paper is as follows. A theoretical portfolio model is
outlined in the next chapter. Thereafter, the empirical model is described
in general terms. The two subsequent chapter deal with the determination of
the exchange rate and of the interest rates, respectively. The seven final
chapters deal with the seven sectors of the model, giving an account of the
demand for and supply of assets by sector.
116 pages, October 1, 1996
Before downloading any of the electronic versions below
you should read our statement on
for viewing Postscript files and the
Acrobat Reader for viewing and printing pdf files.
Full text versions of the paper:
Working-Paper-53-The-Fina ... nometric-Model-Kosmos.pdf
Questions (including download problems) about the papers in this series should be directed to Sarah Hegardt Grant ()
Report other problems with accessing this service to Sune Karlsson ()
or Helena Lundin ().
Design by Joachim Ekebom