Anthony Garratt, Donald Robertson and Stephen Wright, "Permanent vs.
Transitory Components and Economic Fundamentals", Journal of Applied
Econometrics, Vol. 21, No. 4, 2006, pp. 521-542.
There are two files, a program file grw.g and a data file ukmod99.dat.
These are both ASCII files in DOS format. They are zipped in the file
grw-files.zip. Unix users should use "unzip -a".
The results from this paper can be reproduced by running the Gauss
program grw.g. The program will run as it stands, where the data is read
in from the text file ukmod99.dat. Below we describe the data in more
detail. The main output (other files are saved but should be ignored)
from the Guass programe are the *.fmt files:
(1) ov5tr.fmt, which contains the permanent components or trends
(2) ov5cy.fmt, which contains the tranitory component or cycles
Each of these files are n+h+1 by 9 matrices, where n = 140, h=40
(forecast horizon). The permanent and transitory components for output
are contained in column 6 of each matrix respectively, where the numbers
for the period 1965q1-1999q4 are contained in rows 2 to 141 (the first
obervation is 64q4). The other columns contain the permanent and
transitory variables for the remaining variables in the system and are
in the following order (including output, see definitions below):po, rs,
e, r, dpr, y, pps, hy and ys. The estimation period for the model in
the paper is 1965q1-1999q4 (140 observations).
The data file ukmod99.dat is a 148 by 10 matrix containing data for the
period 1963q1-1999q4 (148 observations). The columns contain the
following variables used in the estimation of the model:
Column 1: y, real per capita domestic output
Column 2: ys, real per capita foreign output
Column 3: r, domestic nominal interest rate
Column 4: rs, foreign nominal interest rate
Column 5: e, effective exchange rate
Column 6: hy, real money stock as proportion of real income
Column 7: po, price of oil
Column 8: dpo, change in the price of oil
Column 9: dpr, UK retail price inflation
Column 10: pps, relative prices
The precise definitions and sources of the variables, in the order of the
columns, are:
The definitions and sources of the
variables, in the order given in the file, are:
[1] y: the natural logarithm of UK real per capita domestic output, defined
as
y = ln(GDP/POP),
where GDP is real gross domestic product, at 1995 market prices (index
numbers, 1995=100), seasonally adjusted, source: Office of National
Statistics (ONS) Economic Trends, code YBEZ. POP is total UK population
in thousands, source: ONS, Monthly Digest of Statistics, code DYAY.
[2] ys: the natural logarithm of real per capita foreign output, defined as:
ys = ln(GDP/POP),
where GDP is a total OECD Gross Domestic Product Volume Index
(1995=100), at 1995 market prices, seasonally adjusted, source: OECD,
Main Economic Indicators (MEI), code Q00100319. POP is total OECD
population (adjusted by subtracting the populations of Mexico, Poland,
Hungary and Czech Republic), source: OECD, Labour Force Statistics.
[3] r: the domestic nominal interest rate, measured as a quarterly rate is
computed as:
r = 0.25Wln[1+(R/100)],
where R is the ninety day Treasury Bill average discount rate, at an
annualised rate, source: ONS, Financial Statistics, code AJNB.
[4] rs: the foreign nominal interest rate, measured as a quarterly rate is
computed as:
rs = 0.25Wln[1+( RS/100)],
where RS is a weighted average of foreign annualised interest rates
where the weights are the United States(0.4382), Germany(0.236),
Japan(0.2022) and France(0.1236), taken from the IMFs International
Financial Statistics Yearbook 1998, pages X and Xi. Source: IMFs
International Financial Statistics (IFS). For the US we use the
three-month Treasury Bill rate (IFS Code Q11160C), for Germany the Money
Market Rate (IFS Code Q13460B), for Japan the Money Market Rate (IFS
Code Q15860B) and for France the three month Treasury Bill Rate (IFS
Code Q13260C).
[5] e: the natural logarithm of the UK nominal effective exchange rate is
computed as:
e = -ln(E),
where E is the Sterling Effective Exchange Rate (1995=100, rebased from
1990=100), source: ONS, Financial Statistics, code AJHX.
[6] hy: the natural logarithm of real per capita money stock expressed as a
proportion of real per capita income in computed as:
hy = ln(H/Y),
where H is the M0 definition of the money stock (end period, #Million)
seasonally adjusted, source: ONS, Financial Statistics and Bank of
England. For the period 1969q2-1999q4 we use M0 money stock source: ONS,
Financial Statistics, code AVAE. Nominal income Y is measured using
gross domestic product at market prices (# Million) and is seasonally
adjusted, source: ONS, Economic Trends, code YBHA.
[7] po: the natural logarithm of the oil price is computed as:
po = ln(POIL),
where POIL is the Average Price of Crude Oil, in terms of US Dollars per
Barrel, source: IMF, IFS, code Q00176AAZ, converted into a 1995=100
index.
[8] pps: relative prices defined as:
pps = p -ps
where p is the natural logarithm of the domestic price level and ps is
the natural logarithm of the foreign price index. The domestic price, P,
is measured by the UK Producer Price Index: Output of Manufactured
Products (1995=100), source: ONS, Economic Trends, code PLLU. The
foreign price, PS, is measured by the total OECD Producer Price Index,
1995=100, source: OECD, MEI, code Q005045k. The data used in the
estimation are seasonally adjusted versions of p_{t} or ln(P_{t}), where
the adjustment is performed using the Stamp package (see Harvey,
Koopman, Doornik and Shephard, (1995)).
[9] dpo: the change natural logarithm of the oil price is computed as:
dpo = po - po(-1).
See above.
[10] dpr: the UK inflation rate is computed as:
ln(PR(t) - ln(PR(t-1)),
where PR is the UK Retail Price Index , All Items ( 1995=100, rebased
from 1987=100), source: ONS, Economic Trends, code CHAW. Seasonally
adjusted as above.