Exchange rate forecasting: the errors we’ve really made

We examine the real-time forecasting performance of standard exchange rate models, using dozens of different vintages of data. Favorable evidence of long-horizon exchange rate predictability for the DM and Yen found in Mark (American Economic Review 1995;85:201-218) is present in only a two-year window of data vintages around that originally used. Approximately one-third of the improved forecasting performance over a random walk is eventually undone by data revisions. We also find the models consistently perform better using original release data than fully-revised data, and sometimes forecast better using real-time forecasts of future fundamentals instead of actual future fundamentals, contradicting a cherished presumption dating back to Meese and Rogoff (Journal of International Economics 1983;14:3-24).     

THE STATISTICS OF LONG-HORIZON REGRESSIONS REVISITED1

This paper compares commonly used approaches for estimating the relation between long-horizon returns and a predetermined variable X₁, such as dividend yields. Specifically, we look at regression of (i) nonoverlapping multiperiod returns on X_t (ii) overlapping multiperiod returns on X_t, (iii) single-period returns on multiperiod X_t, and (iv) single-period returns on X_t and its implied long-horizon regression coefficient. We provide analytical formulae which quantify the efficiency of the estimators used in the various approaches. Using the formulae, as well as Monte Carlo simulations, we demonstrate that the relative efficiency of the estimators used in the various approaches differs remarkably, depending on the dynamic structure of the regressor. of special interest for financial economists, when the regressors are highly autocorrelated, we find that the regressions (ii) (iii), and (iv) provide only marginal efficiency gains above and beyond the nonoverlapping long-horizon regression.     

Spillover effects of a mega shopping complex on pre-existing,small retail shops over space,over time and across retail types

This paper studies spillover effects from Starfield, a mega shopping complex in South Korea, to pre-existing, small retail shops over space, over time and across types of retail shops. Using the data on daily store traffics, we test for any complementarity or substitutability from Starfield to neighboring retail shops. Our analysis shows that spillover effects of Starfield on neighboring retail shops are equivocal in the short run. During the first year of the opening of Starfield, the contemporary effect (or the short-lived effect, lasting a single day) of Starfield is negative for consumer goods stores and personal service shops outside the 3-km radius, while it is positive for restaurants and bars within the 5-km radius. After that first year has passed, only the positive contemporary effect on restaurants and bars survives, and the cumulative effect (or the longer-lasting effect) turns significantly positive for consumer goods stores within 1.5 km of Starfield and personal service shops in the 1.5-to-3-kilometer radius. Our analysis estimates the temporal path of spillover effects—both contemporary and cumulative effects—from a newly-built mega shopping complex to neighboring retail shops, which is a novel contribution to the retail literature.     

Do We Fully Understand the Critical Success Factors of Customer Satisfaction with Industrial Goods? - Extending Festge and Schwaiger’s Model to Account for Unobserved Heterogeneity

This paper extends Festge and Schwaiger’s (2007) model of customer satisfaction with industrial goods by accounting for unobserved heterogeneity. The application of a novel response-based segmentation approach in partial least squares path modeling (PLS-PM) - the finite mixture partial least squares (FIMIX-PLS) methodology - opens the way for the effective identification of distinctive customer segments. In comparison to previous studies in this field, group-specific path model estimates reveal each customer segment’s particular characteristics as well as other differentiated findings. Hence, this contribution demonstrates that structural equation modeling studies on the aggregate data level can be seriously misleading and makes a strong case for segment-specific customer satisfaction analyses.     

A Simple Way to Reduce Estimation Bias in Some Dynamic Choice Models

This study proposes a simple estimator for dynamic choice models with continuous unobserved heterogeneity, state dependence induced by a first-order Markov process, and lagged choice coefficients that are not brand specific. As opposed to the estimator most frequently used in marketing studies, which drops the first choice of each panelist, this estimator keeps that choice in the estimation sample but ignores the impact of presample choices on first observed choice. The proposed method presents major advantages over dropping the first observation: (1) it yields low bias estimates for state dependence coefficients and, thus, it can be used to assess the presence and the magnitude of state dependence; (2) it is informative about the magnitude of bias in its estimates; and (3) it yields substantially and significantly smaller biases. The results of this study can be used to determine if the proposed method is suitable for specific applications on real data.     

Fatal violence among children under 15 years in four cities of South Africa, 2001-2005

Homicide rates for South African children were estimated at double the global average in 2000. This article presents a secondary data analysis of the National Injury Mortality Surveillance System (NIMSS), with full coverage in four major metropolitan cities. Child homicide rates for 2001-2005 were calculated within the 0-4, 5-9 and 10-14 year age groups. The homicide rates were similar to the global pattern, with higher rates for boys, and among children aged 0-4 years than for older children. Poisson regression, accounting for city level clustering, was used to investigate age, sex and period effects in the homicide rate. The model indicated that the gender difference was more marked in the 10-14 year age groups (RR?=?2.17; 95% CI 1.97-2.38) than in the 5-9 year (RR?=?1.43; 95% CI 1.27-1.62) with the 0-4 year age group in-between (RR?=?1.80; 95% CI 1.55-2.10). These data confirm previous observations that fatal violence among children is a public health concern, but, given the high rates of homicide among South African adults and other competing public health problems, it is difficult to motivate for action to address the issue of violence against children. Nonetheless, there are sufficient indications that efforts to reduce childhood violence are urgently needed.     

Measuring the benefits of foreign product variety with an accurate variety set

Recent studies have used import data to assess the impact of foreign varieties on domestic prices and welfare. We employ a market-based data set on the U.S. automobile market that allows us to define goods varieties at a more precise level, as well as discern location of production and ownership of varieties. Our estimates of price and welfare changes from new varieties in the U.S. automobile sector are twice as large as standard estimates when using our detailed market-based data. We also show that new varieties introduced by foreign-owned affiliates provided an additional 70% welfare gain during our sample.     

BRAZIL-UNITED STATES TRADE: CAPITAL FLIGHT THROUGH ABNORMAL PRICING

Detailed analysis of the statistics of merchandise trade between Brazil and the United States reveals extensive underpricing of exports and overpricing of imports, which has the effect of transferring large amounts of money out of Brazil and into the United States. Previous studies called attention to this possibility without being able to demonstrate convincingly the extent and amount of the practice. This paper reports the results of a systematic investigation of U.S. customs data at its most disaggregate level to document the amount of capital flows which may be hidden in commodity trade. Using deviations from average prices within commodity classes to identify abnormal prices produces conservative estimates of the amount of capital flight from Brazil to the United States of between $2 to $4 billion in 1995 alone, which would be between 10%-20% of total commodity trade between the countries in that year. Some suggestions are developed for using the results to more closely monitor international transactions in order to reduce this amount.     

我国服务业外商直接投资的经济增长效应——基于9个行业面板数据的实证研究

本文利用固定效应模型对1997-2003年服务业各行业的相关数据进行了短期效应和长期效应的分析,结果表明:服务业外商直接投资具有一定的经济增长效应,但是在不同时间段差异较大。从短期看,1997-2000年期间,服务业外商直接投资具有负的经济增长效应,而在2001-2003年期间,服务业外商直接投资具有正的经济增长效应。从长期看,1997-2000年期间,滞后两年的服务业外商直接投资具有较强的正的经济增长效应,2001-2003年期间,滞后一年的服务业外商直接投资具有比当年更大的正的经济增长效应。最后提出了相关的政策建议。     

Solving and Testing for Regressor-Error (in)Dependence When no Instrumental Variables are Available: With New Evidence for the Effect of Education on Income

This paper has two main contributions. Firstly, we introduce a new approach, the latent instrumental variables (LIV) method, to estimate regression coefficients consistently in a simple linear regression model where regressor-error correlations (endogeneity) are likely to be present. The LIV method utilizes a discrete latent variable model that accounts for dependencies between regressors and the error term. As a result, additional ‘valid’ observed instrumental variables are not required. Furthermore, we propose a specification test based on Hausman (1978) to test for these regressor-error correlations. A simulation study demonstrates that the LIV method yields consistent estimates and the proposed test-statistic has reasonable power over a wide range of regressor-error correlations and several distributions of the instruments. Secondly, the LIV method is used to re-visit the relationship between education and income based on previously published data. Data from three studies are re-analyzed. We examine the effect of education on income, where the variable ‘education’ is potentially endogenous due to omitted ‘ability’ or other causes. In all three applications, we find an upward bias in the OLS estimates of approximately 7%. Our conclusions agree closely with recent results obtained in studies with twins that find an upward bias in OLS of about 10% (Card, 1999). We also show that for each of the three datasets the classical IV estimates for the return to education point to biases in OLS that are not consistent in terms of size and magnitude. Our conclusion is that LIV estimates are preferable to the classical IV estimates in understanding the effects of education on income. JEL Classification: C12, C13, C21, J3, M3