Journal Articles
Discussion Papers
Google+
Twitter
FacebookDiscussion Paper
Abstract
The present value model of the current account has been very popular, as it provides an optimal benchmark to which actual current account series have often been compared. We show why persistence in observed current account data makes the estimated optimal series very sensitive to small-sample estimation error, making it close to impossible to determine whether the paths of the two series truly bear any relation to each other. Moreover, the standard Wald test of the model will falsely accept or reject the model with substantial probability. Monte Carlo simulations and estimations using annual and quarterly data from five OECD countries strongly support our predictions. In particular, we conclude that two important consensus results in the literature – that the optimal series is highly correlated with the actual series, but substantially less volatile – are not statistically robust.
Paper submitted to the special issue "Using Econometrics for Assessing Economic Models" edited by Katarina Juselius.
Data Set
Data sets for articles published in "Economics" are available at Dataverse. Please have a look at our repository.
The data set for this article can be found at: http://hdl.handle.net/1902.1/13900
JEL Classification
Cite As
Assessment
Downloads:
869
Search this article at Google Scholar
Comments and Questions
Here's our full response to the referee report #1
This addendum corrects a simulation contained in our response to the first referee report.
Here's our full response.
see sttached file
Attached is a short note (forthcoming in Applied Economics Letters) that is related to Benoit and Jacques' paper.
Hafedh, thanks for pointing us to your paper, which is highly relevant to and complementary with ours. In particular, it's good to note that both papers find similar results with UK quarterly data. We have resubmitted a revised version of our paper and the new version cites your paper and ...[more]
... discusses your results. Thanks, Jacques
1. The performance of nonlinear Wald tests in small samples may be improved upon by using bootstrap tests.
2. The following drawback of nonlinear Wald test has been noted by Gregory and Veall (Econometrica, 1985). Changing the form of the nonlinear restriction to an algebraically equivalent form under the ...[more]
... null hypothesis changes the numerical value of the Wald test statistic and hence there can be conflicts among different Wald tests. Likelihood ratio and Lagrange Multiplier tests do not suffer from this problem and are therefore preferred.
Sunil,
Thank you for your comments. As you note, there have been previous criticisms of the non-linear Wald test, some of which (but not all of which) we quote in the paper. Remember that our point in the paper is not so much to add to the general/theoretical criticisms ...[more]
... of the non-linear Wald test, but rather to show how the shortcomings of the test apply in this particular literature of present value models of the current account, and how these shortcomings "invalidate" some of the results in this literature. Related to this, we show why the non-linear Wald fails in this particular literature (current account persistence) and show how alternative tests (which we never claim to be "ours") do much better in this particular literature. We believe none of these issues had been signalled before in this particular (but important) literature of intertemporal current accounts. You could say that the problem is that this literature has relied too much on a test which other literatures have shown to be problematic in short samples.
Beyond the issues of the Wald, we believe a key (perhaps the key) contribution of our paper is showing the huge small sample uncertainty surrounding estimates of optimal current accounts. In the end, we believe the reason the model has been so popular is because of its supposed ability to generate optimal series which can be compared with the actual. As we show, these comparisons are much less useful than people have made them to be.
Thanks for your comments