Front Matter Page European Department
I. Introducing Optimal Pooling with Machine Learning
II. A Two-Step Method for Optimal Pooling
III. Applying the Method
II. IV. Conclusions
1. The Bias-Variance Tradeoff
1. Step 1—Proximity
2. Step 2—Optimal Pool
3. Turkey-Most ProximateCountries
4. Turkey-Least ProximateCountries
5. Turkey—Relative Forecast Errors for Different Pools
6. Other Countries—Relative Forecast Error of Different Pools
I. Machine Learning and Cross
We leverage insights from machine learning to optimize the tradeoff between bias and variance when estimating economic models using pooled datasets. Specifically, we develop a simple algorithm that estimates the similarity of economic structures across countries and selects the optimal pool of countries to maximize out-of-sample prediction accuracy of a model. We apply the new alogrithm by nowcasting output growth with a panel of 102 countries and are able to significantly improve forecast accuracy relative to alternative pools. The algortihm improves nowcast performance for advanced economies, as well as emerging market and developing economies, suggesting that machine learning techniques using pooled data could be an important macro tool for many countries.
output growth with a panel of (up to) 102 countries for the period 1987–2018. The nine example countries are Austria, Canada, Costa Rica, El Salvador, Germany, Lithuania, Mexico, Iceland and Turkey. For details on indicator selection, variable transformations, and missing value imputation, see Annex III .
In determining proximity, the algorithm selects countries that are similar in terms of economic structure . Figures 3 and 4 plot the most and least proximatecountries for Turkey. In this case, the most proximatecountries include emerging market countries with
costs should have augmented trade volumes.
This predicted negative EU effect, however, is well understood in the literature (see e.g., Linnemann, 1966 ; Aitken, 1973 ; Pollak, 1996 ; Rose, 2004 ; Baldwin 2006). Dating back to Linnemann (1966) , the gravity equation has been known to systematically over-predict trade among large, geographically proximatecountry pairs. Europe-specific variables thus tend to pick up the negative residuals resulting from proximate European countries’ under-trading relative to gravity model predictions. Since the EU variable most
The introduction of the euro generated substantial interest in measuring the impact of currency unions (CUs) on trade flows. Rose's (2000) initial estimates suggested a tripling of trade and created a literature in search of "more reasonable" CU effects. A recent meta-analysis of this literature shows that subsequent papers quantify CU trade impacts at 30-90 percent. However, most recent studies use shorter time series and fewer countries than Rose in his original work. We revisit Rose's original benchmark, extend the dataset, and address Baldwin's (2006) critiques regarding the proper specification of gravity models in large panels by simultaneously accounting for multilateral resistance and unobserved bilateral heterogeneity. This produces a robust average CU trade effect of 45 percent. Yet, the trade impacts of individual CUs vary substantially and are generally lower than those of preferential trade agreements (PTAs). Our revised benchmark can be used as a yardstick for future studies to delineate how estimates differ due to new data or differences in econometric specifications.
Mr. Chris Papageorgiou, Christian Henn, and Theo S. Eicher
costs (see Freund, 2000 ). Pollak (1996) points out that it has been well known since the original Linnemann (1966) gravity specification that the approach systematically over-predicts trade among geographically proximatecountries and under-predicts trade between distant country pairs. Gravity model refinements have attempted to capture some of this effect by adding the Remoteness variable. The comparison between our specification 1 and 2 shows, however, that this does not purge the entire systematic error. In the presence of systematic under-trading compared
Mr. Chris Papageorgiou, Christian Henn, and Theo S. Eicher
Trade theories covering Preferential Trade Agreements (PTAs) are as diverse as the literature in search of their empirical support. To account for the model uncertainty that surrounds the validity of the competing PTA theories, we introduce Bayesian Model Averaging (BMA) to the PTA literature. BMA minimizes the sum of Type I and Type II error, the mean squared error, and generates predictive distributions with optimal predictive performance. Once model uncertainty is addressed as part of the empirical strategy, we report clear evidence of Trade Creation, Trade Diversion, and Open Bloc effects. After controlling for natural trading partner effects, Trade Creation is weaker - except for the EU. To calculate the actual effects of PTAs on trade flows we show that the analysis must be comprehensive: it must control for Trade Creation and Diversion as well as all possible PTAs. Several prominent control variables are also shown to be robustly related to Trade Creation; they relate to factor endowments and economic policy.
—to countries to which they either are politically proximate or have strong business links. Results from adding a large set of political economy variables to the baseline frequentist specifications are presented in tables I.6 and I.7. No consistent patterns for the importance of political economy variables can be found, and the vast majority of them are demonstrated utterly insignificant in explaining program design. Where they do enter significantly, they generally carry counterintuitive signs, suggesting harsher conditionality for proximatecountries. In a few cases, the
Design of Fund-supported programs aims to address country specific needs while remaining even-handed and consistent with Fund policy. This paper examines the extent to which program design and conditionality have been appropriate in pursuing these goals, by seeking to answer several questions: has program design been consistent and evenhanded; has it addressed country specific needs and objectives appropriately; has it been based on reasonably good macroeconomic projections; and has it been flexible in the face of evolving country circumstances. The description and analysis focuses on the period between 2006 and September 2011, with some attention to the 2002-05 period.
-determined distribution of exports and imports. In particular, the geographically proximatecountries of the European Union now play a much more prominent role. The increasing reintegration into the global trading system has already been accompanied by some growth in productivity and wages in the transition countries, and these trade links represent an important channel through which these countries are gaining technological knowledge and managerial skills. Even though trade restrictions in the advanced economies do not in general appear to form a major impediment to the export of most