Modelling and comparing OECD countries' consumer behaviour.
This thesis seeks to model and compare OECD countries' consumer behaviour. We build
REPIHIRELCH and ECM formulations using long time-series based solely upon private sector
measures of income for twenty OECD countries. No previous study features such a broad
coverage of private sector data and models.
Using the Johansen procedure we build structural ECMs based upon consumption, income and
inflation allowing for heterogeneous dynamics across countries and considering whether an
intercept should be included in, or excluded from, the cointegrating vector. Models embodying
asymmetric nonlinear adjustment towards equilibrium are also developed. We are not aware of
any previous study which considers such flexibility of specification for twenty OECD
economies. \Ve build ECMs consistent with valid error-correction behaviour for eighteen
countries and find evidence favouring asymmetric/nonlinear adjustment for twelve countries.
We derive a REPIHfRELCH model in logarithmic form to allow for current income consumers,
durable expenditures and intertemporal substitution. We are aware of no previous study which
simultaneously allows for all three of these features in a REPIHIRELCH model. This model is
estimated with both GMM and IV methods. A proportion of current income consumers is found
for all twenty countries and, in addition, accommodation for durability is evident for two
economies. There is no evidence of intertemporal substitution.
Regressions are employed to explain the cross-country variations in the models' estimated
parameters. We are not aware of any previous study attempting to explain variations in estimated
elasticities from an ECM. We are able to explain the cross-country variations in the long run
income and inflation elasticities but not the short run income and inflation elasticities or the
adjustment coefficient. Only one previous study considers whether the cross-country variation
in the estimated proportion of current income consumers can be explained by liquidity
constraints. We use a broader range of proxies for liquidity constraints and additionally consider
income uncertainty as a potential explanation. Application of iterative NL3 SLS to the whole
panel reveals that both factors explain the cross-country variation in this proportion.