Search

Using Monte Carlo experiments, we examine the performance of indirect inference tests of DSGE models in small samples, using various models in widespread use. We compare these with tests based on direct inference (using the Likelihood Ratio). We find that both tests have power so that a substantially false model will tend to be rejected by both; but that the power of the indirect inference test is by far the greater, necessitating re-estimation to ensure that the model is tested in its fullest sense. We also find that the small-sample bias with indirect estimation is around half of that with maximum likelihood estimation.

With Monte Carlo experiments on models in widespread use we examine the performance of indirect inference (II) tests of DSGE models in small samples. We compare these tests with ones based on direct inference (using the Likelihood Ratio, LR). We find that both tests have power so that a substantially false model will tend to be rejected by both; but that the power of the II test is substantially greater, both because the LR is applied after reestimation of the model error processes and because the II test uses the false model.s own restricted distribution for the auxiliary model's coefficients. This greater power allows users to focus this test more narrowly on features of interest, trading off power against tractability.

We extend the method of indirect inference testing to data that is not filtered and so may be non-stationary. We apply the method to an open economy real business cycle model on UK data. We review the method using a Monte Carlo experiment and find that it performs accurately and has good power.

Using Monte Carlo experiments, we examine the performance of Indirect Inference tests of DSGE models, usually versions of the Smets-Wouters New Keynesian model of the US postwar period. We compare these with tests based on direct inference (using the Likelihood Ratio), and on the Del Negro-Schorfheide DSGE-VAR weight. We find that the power of all three tests is substantial so that a false model will tend to be rejected by all three; but that the power of the indirect inference tests are by far the greatest, necessitating re-estimation by indirect inference to ensure that the model is tested in its fullest sense.

Using Monte Carlo experiments, we examine the performance of Indirect Inference tests of DSGE models, usually versions of the Smets-Wouters New Keynesian model of the US postwar period. We compare these with tests based on direct inference (using the Likelihood Ratio), and on the Del Negro-Schorfheide DSGE-VAR weight. We find that the power of all three tests is substantial so that a false model will tend to be rejected by all three; but that the power of the indirect inference tests are by far the greatest, necessitating re-estimation by indirect inference to ensure that the model is tested in its fullest sense.

We review the methods used in many papers to evaluate DSGE models by comparing their simulated moments and other features with data equivalents. We note that they select, scale and characterise the shocks without reference to the data; crucially they fail to use the joint distribution of the features under comparison. We illustrate this point by recomputing an assessment of a two-country model in a recent paper; we find that the paper's conclusions are essentially reversed.

We evaluate the Smets-Wouters model of the US dynamically using indirect inference with a VAR representation of the main US data series. We find that the New Keynesian SW model is badly rejected by the data's dynamic properties and in particular cannot match the variability of the data. An alternative (New Classical) version of the model with flexible wages and prices and a one-period information lag fares no better. A weighted model (mostly NC but part NK) is able to match the data variability, though it too is rejected overall. Allowing for structural breaks in monetary regime we find a model from 1984 onwards fits fairly well dynamically; this has a high NK weight, suggesting much greater nominal stickiness during the great moderation.

We evaluate the Smets-Wouters model of the US using indirect inference with a VAR representation of the main US data series. We find that the original New Keynesian SW model is on the margin of acceptance when SW's own estimates of the variances and time-series behaviour of the structural errors are used. However when the structural errors implied jointly by the data and the structural model are used the model is rejected. We also construct an alternative (New Classical) version of the model with flexible wages and prices and a one-period information lag. This too is rejected. But when small proportions of both the labour and product markets are assumed to be imperfectly competitive within otherwise flexible markets the resulting `weighted' model is accepted.

We review the methods used in many papers to evaluate DSGE models by comparing their simulated moments with data moments. We compare these with the method of Indirect Inference to which they are closely related. We illustrate the comparison with contrasting assessments of a two-country model in two recent papers. We conclude that Indirect Inference is the proper end point of the puzzles methodology.

We examine a two country model of the EU and the US. Each has a small sector of the labour and product markets in which there is wage/price rigidity, but otherwise enjoys flexible wages and prices with a one quarter information lag. Using a VAR to represent the data, we find the model as a whole is rejected. However it is accepted for real variables, output and the real exchange rate, suggesting mis-specification lies in monetary relationships. The model highlights a lack of spillovers between the US and the EU.

We examine a two country model of the EU and the US. Each has a small sector of the labour and product markets in which there is wage/price rigidity, but otherwise enjoys flexible wages and prices with a one quarter information lag. Using a VAR to represent the data, we find the model as a whole is rejected. However it is accepted for real variables, output and the real exchange rate, suggesting mis-specification lies in monetary relationships. The model highlights a lack of spillovers between the US and the EU.

We review the methods used in many papers to evaluate DSGE models by comparing their simulated moments and other features with data equivalents. We note that they select, scale and characterise the shocks without reference to the data; crucially they fail to use the joint distribution of the features under comparison. We illustrate this point by recomputing an assessment of a two-country model in a recent paper; we find that the paper's conclusions are essentially reversed.

We review the methods used in many papers to evaluate DSGE models by comparing their simulated moments with data moments. We compare these with the method of Indirect Inference to which they are closely related. We illustrate the comparison with contrasting assessments of a two-country model in two recent papers. We conclude that Indirect Inference is the proper end point of the puzzles methodology.