Robert B. Gramacy, Samuel W. Malone, and Enrique ter Horst, "Exchange Rate Fundamentals, Forecasting, and Speculation: Bayesian Models in Black Markets", Journal of Applied Econometrics, Vol. 29, No. 1, 2014, pp. 22-41. The data are in the file gmh-data.txt, an ASCII file in DOS format which is zipped in the file gmh-data-text.zip. Unix/Linux users should use "unzip -a". The same data are also in a Stata dataset gmh-data.dta, which is zipped in the file gmh-data-dta.zip. The data are organized with variables in columns, with successive rows within a country representing months, and countries stacked on top of each other vertically in alphabetical order by country name. There are 34 countries in total. The list of countries and the number of observations by country is given below: Country name Number of observations ARGENTINA 60 BANGLADESH 60 CHILE 60 BOLIVIA 36 COSTA RICA 60 EGYPT 211 EL SALVADOR 144 GHANA 57 ICELAND 48 INDIA 324 INDONESIA 108 IRELAND 60 JAMAICA 60 JORDAN 127 KENYA 282 MALAYSIA 276 MOROCCO 106 MYANMAR 135 NEPAL 60 NIGERIA 113 PAKISTAN 312 PARAGUAY 103 PHILLIPPINES 55 SRI LANKA 144 TAIWAN 90 THAILAND 156 TURKEY 60 UGANDA 60 URUGUAY 48 COLOMBIA 36 GAMBIA, THE 60 SOUTH AFRICA 247 SOUTH KOREA 120 VENEZUELA 78 The above information can be gleaned in the Stata version of the dataset by executing the command by cntry: summarize LNparallelrateindex Some of the variables included in the file are instrumental for calculating other variables, but are not used directly in the forecasting models themselves. The variables incldued in the dataset file are as follows: --- Variable name Description Source country The country name Authors date The date code Authors parallelrate The black market exchange rate Reinhart and Rogoff (2004) officialrate The official exchange rate Reinhart and Rogoff (2004) parallelrateindex The normalized black market rate Authors officialrateindex The normalized official rate Authors cpi_loc The local currency CPI IMF cpi_us The US CPI IMF i_loc The local currency interest rate as an APR IMF i_usd The US interest rate as an APR IMF roer The real official exchange rate Authors r_usd The monthly US interest rate (i_usd/12) Authors r_loc The monthly local currency interest rate (i_loc/12) Authors d The monthly growth rate of the parallel rate index Authors Rcomplete The USD return on a reverse carry trade Authors R Rcomplete during out-of-sample months only Authors DREP Calculated as R*(1+d), following Fishelson (1988) Authors LNparallelrateindex The natural log of parallelrateindex Authors LNcommodity The natural log of commodity Authors LNroer The natural log of roer Authors LNc the logarithm of the ratio of M2 to officialrate Authors and IMF for M2 LNreserves The natural log of foreign reserves Authors and IMF for Reserves DLNroer The first difference of LNroer Authors DLNc The first difference of LNc Authors DLNreserves The first difference of LNreserves Authors DLNcommodity The first difference of LNcommodity Authors DDREP The first difference of DREP Authors LNofficialrateindex The natural log of officialrateindex Authors DLNofficialrateindex The first difference of LNofficialrateindex Authors DLNparallelrateindex The first difference of LNparallelrateindex Authors cntry The STATA country code for country Authors commodity Country-specific commodity price index: the price of Authors calculations and United Nations Commodity the country's primary export commodity during the Trade Statistics for commodity prices by country black market episode. --- Users should note that, in calculating (reverse) carry trade returns, we use the formulas Rcomplete = 1 + r_usd - (1 + r_loc) / (1 + d) and d_{t+1} = E_{t+1}/E_{t} - 1 where E = parallelrateindex. The variable DREP is computed in an analogous manner, as described in the text and above.