Challenging the Empirical Evidence From Present Value Models of the Current Account

IMF Working Papers, June 1, 2004

Source details

Canonical URL: Challenging the Empirical Evidence From Present Value Models of the Current Account

Under near-singularity conditions typically generated by persistence in current account data the predictions of present value models become extremely sensitive to small sample estimation error.
Traditional Wald tests will distort the likelihood that the model is true.
Using OECD data the authors find:
- (i) the Wald test often leads to the wrong inference compared to a valid test;
- (ii) in all cases posterior distributions of the predicted series and associated correlation coefficients and variance ratios are very wide.
In particular, one cannot draw any firm conclusion regarding excess current account volatility.

Topic areas: Balance of payments, Current account, Econometric analysis, National accounts, Personal income, Vector autoregression.
Keywords preserved from the source: correlation coefficient, Current account, current account persistence, current account series, current account volatility, model evaluation, Personal income, predicted current account, present value model, VAR companion matrix, VAR parameter estimate, variance-covariance matrix, Vector autoregression, WP.
The analysis centers on how persistence in current account series generates near-singularity conditions that amplify sensitivity to small sample estimation error in present value models.

Small sample estimation error under near-singularity undermines robustness of model predictions.
Conventional inference using Wald tests is shown to be unreliable in this context; comparisons to a valid test reveal frequent incorrect inferences from the Wald approach.
Wide posterior distributions for predicted series, correlation coefficients, and variance ratios indicate large uncertainty in estimates and inferences drawn from present value models applied to persistent current account data.

Empirical evidence from present value models of the current account is fragile when current account data exhibit persistence that induces near-singularity.
Given the wide posterior distributions and the propensity for standard Wald tests to mislead, researchers and policymakers should be cautious in interpreting results from present value models regarding excess current account volatility.
The paper implies the need for valid testing procedures and an acknowledgment of large estimation uncertainty when assessing present value model predictions for current accounts.