Knowledge that Transforms

It is often documented, based on autocorrelation, variance ratio, and power spectrum, that exchange rates approximately follow a martingale process. Because these data check serial uncorrelatedness rather than martingale difference, they may deliver misleading conclusions in favor of the martingale hypothesis when the test statistics are insignificant. In this paper, we explore whether there exists a gap between serial uncorrelatedness and martingale difference for exchange rate changes, and if so, whether nonlinear time series models admissible in the gap can outperform the martingale model in out-of-sample forecasts. Applying the generalized spectral tests of Hong to five major currencies, we find that the changes of exchange rates are often serially uncorrelated, but there exists strong nonlinearity in conditional mean, in addition to the well-known volatility clustering. To forecast the conditional mean, we consider the linear autoregressive, autoregressive polynomial, artificial neural network, and functional-coefficient models, as well as their combination. The functional coefficient model allows the autoregressive coefficients to depend on investment positions via a moving-average technical trading rule. We evaluate out-of-sample forecasts of these models relative to the martingale model, using four criteria—the mean squared forecast error, the mean absolute forecast error, the mean forecast trading return, and the mean correct forecast direction. White's reality check method is used to avoid data-snooping bias. It is found that suitable nonlinear models, particularly in combination, do have superior predictive ability over the martingale model for some currencies in terms of certain forecast evaluation criteria.

This paper provides empirical evidence showing the effect of airline cooperation on the interline fares paid by international passengers. The analysis focuses on two measures of cooperation, codesharing and antitrust immunity, and the results show that their partial effects are both negative. The presence of codesharing on an international interline itinerary reduces the fare by 8%–17%, with the exact number depending on the sample used and the estimation method. Moreover, the presence of antitrust immunity reduces the fare by 13%–21%. Codesharing and immunity are substitutes, however, in the sense that their combined effect is smaller than the sum of their partial effects. Taking account of this difference, which is captured by an interaction variable in the regressions, the combined effect ranges between 17% and 30%. These results provide strong evidence that airline cooperation in the fare-setting process generates substantial benefits for interline passengers.

Environmental regulation in the United States has undergone a slow evolution from command and control strategies towards market-based regulations. One such innovation is the Toxics Release Inventory (TRI), a regulation that requires polluting firms to publicly disclose information about their toxic emissions. The basic tenet of this regulation is that it corrects for informational asymmetries between polluters and households, allowing communities to pressure polluters to decrease their emissions. Policy-makers have judged the TRI a tremendous success, as national releases declined by 43% between 1988 and 1999. Yet many of the fundamental problems which are known to lead to the classic failure of the Coase theorem (such as high transaction costs and difficulties in organizing) cast doubt on the effectiveness of disclosure rules, alone, to lead to an efficient outcome in the case of pollution. We use an event study methodology with high-quality data on house prices and other local attributes to assess the extent to which the public values changes in toxic releases and thus the success of TRI. Our major findings include: (1) declines in toxic releases appear unrelated to any political economy variables that might lead to public activism; (2) initial information released under TRI had no significant effect on the distribution of house prices; and (3) house prices show no significant impact of declines in reported toxic releases over time. Standard errors are small enough that we can reject the hypothesis that large declines in toxic releases lead to more than a 0.5% increase in house prices. These results also hold when we control for differences in the availability of information on TRI and the possible effect of expectations. Our findings cast doubt on the ability of the public to process complex information on hazardous emissions and support the Coase theorem in that right-to-know laws such as TRI may not be the most effective form of environmental regulation.

This paper uses a real-time data set to analyze data revisions and to test the robustness of published econometric results. The data set consists of vintages, or snapshots, of the major macroeconomic data available at quarterly intervals in real time. The paper illustrates why such data may matter, examines the properties of several of the variables in the data set across vintages, and examines key empirical papers in macroeconomics, investigating their robustness to different vintages.

The Gini coefficient is a downward-biased measure of inequality in small populations when income is generated by one of three common distributions. The paper discusses the sources of bias and argues that this property is far more general. This has implications for (i) the comparison of inequality among subsamples, some of which may be small, and (ii) the use of the Gini in measuring firm size inequality in markets with a small number of firms. The small-sample bias has often led to misperceptions about trends in industry concentration. A small-sample adjustment results in a reduced bias, which can no longer be signed. This remaining bias rises with the dispersion and falls with increasing skewness of the distribution. Finally, an empirical example illustrates the importance of using the adjusted Gini. In this example it is shown that, controlling for market characteristics, larger shipping cartels include a set of firms that is stochastically identical (in terms of relative size) to those of smaller shipping cartels.

We distinguish between three different strategies for estimating forecasting equations with real-time data and argue that the most popular approach should generally be avoided. The point is illustrated with a model that uses current-quarter monthly industrial production, employment, and retail sales data to predict real GDP growth. When the model is estimated using either of our two alternative methods, its out-of-sample forecasting performance is superior to that obtained using conventional estimation and compares favorably with that of the Blue Chip consensus.

This study uses survey data of fund managers' views on prospects for international equity markets to shed light on why investment portfolios are significantly biased towards domestic equities. We find that fund managers from the United States, the United Kingdom, continental Europe, and Japan show a significant relative optimism towards their home equity market. Where institutional factors have largely failed to explain the puzzle, our evidence lends support to behavioral explanations of the bias.

This paper offers a new approach to estimating time-varying covariance matrices in the framework of the diagonal-vech version of the multivariate GARCH(1,1) model. Our method is numerically feasible for large-scale problems, produces positive semidefinite conditional covariance matrices, and does not impose unrealistic a priori restrictions. We provide an empirical application in the context of international stock markets, comparing the new estimator with a number of existing ones.

This paper proposes a nonparametric test for parametric conditional distributions of dynamic models. The test is of the Kolmogorov type coupled with Khmaladze's martingale transformation. It is asymptotically distribution-free and has nontrivial power against root-n local alternatives. The method is applicable for various dynamic models, including autoregressive and moving average models, generalized autoregressive conditional heteroskedasticity (GARCH), integrated GARCH, and general nonlinear time series regressions. The method is also applicable for cross-sectional models. Finally, we apply the procedure to testing conditional normality and the conditional t-distribution in a GARCH model for the NYSE equal-weighted returns.

In a cross-country data set for developing and transitional economies, private consumption per capita from the national accounts deviates on average from mean household income or expenditure based on national sample surveys. Growth rates also differ systematically, so that the ratio of the survey mean to mean consumption from the national accounts tends to fall over time. The exceptions to these general findings are revealing, however. There are strong regional effects. The aggregate difference in the levels is due more to income surveys than to expenditure surveys. Divergence over time is mainly due to the severe data problems in the (contracting) transition economies.