Huixin Bi and Nora Traum, "Estimating Fiscal Limits: The Case of Greece," Journal of Applied Econometrics, Vol. 29, No. 7, 2014, pp. 1053-1072. The file BiTraum_Appendix.pdf contains an online appendix. The data used for estimation are contained in BiTraum_data.txt, an ASCII file in DOS format that is zipped in bt-data.zip. Unix/Linux users should use "unzip -a". The file contains 40 observations (in rows) for 5 data series (in columns). The series are organized as follows: real output (Y), government spending to GDP ratio (G), tax revenue to GDP ratio (TR), government debt to GDP ratio (B), and the 10-year real interest rate (R). Five observables for Greece over the period 2001Q1-2010Q4 are used for the estimation: real output, government spending to GDP ratio, tax revenue to GDP ratio, government debt to GDP ratio, and the 10-year real interest rate. Thus, we have 40 observations for each 5 data series. Details of how each data series were constructed are as follows: Real GDP. Constructed by dividing the nominal quarterly gross domestic product from the OECD quarterly National Accounts (using the expenditure approach, series B1 GE) by the gross domestic product deflator (constructed using the expenditure approach, series B13). Real Gov. Spending. Constructed using general government final consumption expenditure from the OECD quarterly National Accounts (series P3S13) divided by the gross domestic product deflator (constructed using the expenditure approach, series B13). Real Tax Revenue. A quarterly tax revenue series is not provided by the OECD. Thus, we construct a quarterly measure in the following way. First, we construct a measure of total tax revenue by combining Eurostat quarterly series for tax receipts on income/wealth, production and imports, capital taxes, and social contributions. We seasonally adjust this series using Demetra+ and the tramo-seat RSA4 specification. Next, using an annual nominal tax revenue series from the OECD volume 90 (consisting of indirect and direct taxes and social security contributions, TIND + TY + SSRG), we interpolate a quarterly frequency series using the method of Chow and Lin (1971) and the seasonally adjusted quarterly Eurostat tax revenue series for the interpolation. Finally, we construct a quarterly real tax revenue series by dividing the interpolated series by the OECD’s gross domestic product deflator (constructed using the expenditure approach, series B13). Real Gov. Debt. A quarterly government debt series is not provided by the OECD. Thus, we construct a quarterly measure in the following way. First, we seasonally adjust using Demetra+ and the tramo-seat RSA4 specification the Eurostat quarterly series for nominal gross government consolidated debt. Next, using the annual nominal gross public debt series (under the Maastricht criterion) from the OECD volume 90, we interpolate a quarterly frequency series using the method of Chow and Lin (1971) and the seasonally adjusted quarterly Eurostat debt series for the interpolation. Finally, we construct a quarterly real debt series by dividing the interpolated series by the OECD’s gross domestic product deflator (constructed using the expenditure approach, series B13). Real Interest Rate. To construct a 10-year real interest rate measure, we use data for the nominal interest rate, (taken from the BIS) and the expected inflation rate, πe. Our measure of expected inflation for Greece is the expected inflation series from the Survey of Professional Forecasts EU-area five year ahead expected inflation series. The gross real interest rate is constructed using the relation, R = (1 + i)/(1+πe), where R stands for the gross real interest rate, i stands for the nominal interest rate, and πe is our measure of inflation expectations. We calculate the government spending to GDP ratio, tax revenue to GDP ratio, and government debt to GDP ratio by taking each real fiscal series described above and dividing by our real GDP series. The estimation uses these three series, along with the real GDP and real interest rate series described above. For each series, we transform the series into percentage deviations from its mean value over the period 2001Q1-2010Q4. In addition, the real GDP series is linearly detrended.