Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Forecasting Performance of Seasonal Cointegration Models: Marten Lof, Johan Lyhagen

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

International Journal of Forecasting 18 (2002) 31–44

www.elsevier.com / locate / ijforecast

Forecasting performance of seasonal cointegration models


˚
Marten ¨ *, Johan Lyhagen
Lof
Department of Economic Statistics, Stockholm School of Economics, P.O Box 6501 SE-113 83 Stockholm, Sweden

Abstract

Forecasts from two different seasonal cointegration specifications are compared in an empirical forecasting example and in
a Monte Carlo study. The two seasonal cointegration specifications are the one proposed by Lee [Journal of Econometrics 54
(1992) 1], with a parameter restriction included at the annual frequency, and the model proposed by Johansen and
Schaumburg [Journal of Econometrics 88 (1998) 301], with a general specification for the complex root frequency,
respectively. In the empirical forecasting example we also include a standard cointegration model based on first differences
and seasonal dummies and analyze the effects of restricting or not restricting seasonal dummies in the seasonal cointegration
models. While the Monte Carlo results favor the specification suggested by Johansen and Schaumburg, and definitely so if
larger sample sizes are considered, we do not find such clear cut evidence in the empirical example.  2002 International
Institute of Forecasters. Published by Elsevier Science B.V.

Keywords: Seasonal cointegration; Monte Carlo

1. Introduction about short and long run dynamics in the


system. These negative effects may be serious
It is common in applied work to assume an when the model is used in forecasting.
approximately constant seasonal pattern, mod- Similar to the approach suggested by Johan-
elled by seasonal dummies. Recently the unit sen (1988) and Johansen (1995) for the zero
root and cointegration analysis at the zero frequency, Lee (1992) presents a maximum
frequency have been extended to seasonal fre- likelihood estimator for seasonally cointegrating
quencies. Nonzero frequency unit roots imply a relations. The seasonal error correction model
nonstationary seasonal pattern. Assuming a [SECM], which allows for stochastic seasonality
deterministic seasonal pattern in a multivariate is, however, only partially correct when it
setting, when it is in fact stochastically evolving comes to testing for the cointegrating rank at the
over time, could lead to inappropriate inference annual frequency. A certain parameter restric-
tion has been used in the literature to overcome
* Corresponding author. Tel.: 146-8-736-9221; fax: this problem. If this restriction is relaxed one
146-8-348-161. cannot apply estimation methods based on
¨
E-mail address: stmlo@hhs.se (M. Lof). canonical correlations, as suggested by Lee.

0169-2070 / 02 / $ – see front matter  2002 International Institute of Forecasters. Published by Elsevier Science B.V.
PII: S0169-2070( 01 )00105-4
32 ¨ , J. Lyhagen / International Journal of Forecasting 18 (2002) 31 – 44
M. Lof

Johansen and Schaumburg (1998) argue that casting effects of first deleting then either
this is a peculiar restriction as it constrains all restricting or not restricting the seasonal inter-
coefficients at the annual frequency. They pres- cepts as discussed above. Using three empirical
ent another estimation procedure for the param- data sets and various forecasting periods they
eters, corresponding to the annual frequency and show that the suggested restricted seasonal
introduce a general asymptotic theory for the dummy approach yields better forecasts in most
seasonal cointegration model. cases.
Kunst and Franses (1998) argue that deter- In the present study we show results from an
ministic seasonal dummy variables, which are empirical forecasting example, but we also
often included unrestrictedly in the SECM, conduct a Monte Carlo study where seasonally
should be confined to the seasonal cointegrating cointegrated data generating processes
relations instead. If cointegration at seasonal [DGPs], with different parameter constellations,
frequencies exists, an inclusion of unrestricted are investigated. In the empirical forecasting
seasonal intercepts then implies a growing part we use macroeconomic data sets for Au-
amplitude in the seasonal cycles, which may not stria, Germany and the United Kingdom, which
be a realistic assumption in most cases. are also used in Kunst and Franses (1998). In
In the present study, various seasonal coin- the Monte Carlo study we analyze systems
tegration specifications will be contrasted to the including three variables, whereas six variables
model suggested by Johansen and Schaumburg are used in the empirical example. The forecast-
(1998) [henceforth JS]. We have found three ing performance of the more general specifica-
studies on forecasting and seasonal cointegra- tion for the annual frequency, recently proposed
tion: Kunst (1993a); Reimers (1997) and Kunst by JS is evaluated. We compare that specifica-
and Franses (1998). Kunst (1993a) contrasts the tion with the original version of the seasonal
SECM, where an intercept is included and the cointegration model, proposed by Lee (1992).
restriction on the annual frequency is imposed, In the empirical forecasting example we also
to a vector error correction model [VECM] and compare these two seasonal cointegration spe-
also a VAR model in first differences and cifications with a VECM in first differences.
seasonal dummies. Two examples based on real The remainder of this paper is organized as
data and a Monte Carlo experiment show that follows. Section 2 presents specifications of the
the benefits from accounting for seasonal coin- SECM, while Section 3 describes the data and
tegration are quite limited. Reimers (1997) uses the estimation results for the models to be used
the conventional seasonal cointegration model in the empirical forecasting example. Section 4
suggested by Lee, with another simplifying presents forecast performance when using the
restriction on the annual frequency and consi- empirical data sets. In Section 5 the Monte
ders a simulated two variable seasonal coin- Carlo setup is discussed. A comparison of
tegration model, with a fixed lag length and no model forecasts from the Monte Carlo study is
seasonal intercepts. The SECM is compared to a carried out in Section 6. The final section
traditional VECM in first differences with contains conclusions.
seasonal dummies. The main conclusion is that
models in first differences produce smaller
forecast errors for short horizons, but when 2. Seasonal cointegration
longer forecasting periods are considered the
seasonal cointegration model appears preferable. Lee (1992) suggests a maximum likelihood
Kunst and Franses (1998) investigate the fore- estimator for seasonal cointegration relations.
¨ , J. Lyhagen / International Journal of Forecasting 18 (2002) 31 – 44
M. Lof 33

This procedure extends the maximum likelihood cycles. If this restriction [henceforth denoted as
approach, which is summarized in Johansen the annual restriction] is relaxed one cannot
(1995). Assuming quarterly data and that D4 Yt apply the estimation method that uses canonical
is stationary, a seasonal VECM of order p of the correlations. There is also some evidence in the
following form is considered: literature that the absence of non-synchronous
p seasonal cycles should have little effect on the
O a b 9 Z 1O G D Y
3

Dn 4 Yt 5 1 F Dt 1 ´t, test for cointegration at frequency p / 2, see Lee


i i i,t j 4 t2j
i51 j 51 (1992).
(1) JS, who argue that the above mentioned
restriction is peculiar and not justified from a
where the Dt are deterministic components and theoretical point of view, refine the theory for
where ´t is i.i.d. Nn (0,V ). The process D4 Yt seasonal cointegration in the general case. They
above is said to be seasonally cointegrated if propose the following error correction model for
and only if at least one of the ai b 9i matrices for quarterly data:
i 5 2,3 on the right hand side has reduced,
Oa b 9X
2
non-zero rank. The linear filters which in (1)
remove all unit roots except those on the zero, D4 Yt 5 i i i,t 1 2(aR b R9 1 aI b I9 )XR,t
i51
biannual and annual frequencies, respectively 1 2(aR b I9 2 aI b 9R )XI,t
are: p
2
Z1,t 5 (L 1 L 1 L 1 L )Yt ,
2 3
3 4

4
1 OG D Y
j 51
j 4 t2j 1 F Dt 1 ´t , (2)
Z2,t 5 (L 2 L 1 L 2 L )Yt ,
where the processes X1,t , . . . ,XI,t are:
Z3,t 5 (L 2 2 L 4 )Yt .
1
These filters are the vector equivalents of the X1,t 5 ]Z1,t ,
4
univariate transformations used in the so-called 1
HEGY-test for seasonal unit roots, see Hy- X2,t 5 2 ]Z2,t ,
4
lleberg, Engle, Granger and Yoo (1990) 1
[HEGY]. Furthermore, the regressors Zi,t are XR,t 5 ]Z3,t ,
4
asymptotically uncorrelated: 1
XI,t 5 2 ]Z4,t ,
4
OZ Z 9 → 0,i ± j,
T
22
T i,t j,t P respectively. It can be seen that the annual
t 51
restriction would imply that aR b 9I 2 aI b R9 5 0
implying that the cointegration vectors and in (2), which is a strong restriction on the
adjustment coefficients can be found by remov- coefficients at the complex root frequency.
ing the reduced rank restriction on the other The estimation of ai b 9i for i 5 1,2 uses
frequencies by concentrating out the associated canonical correlations in analogy to the Johan-
regressors. sen procedure and hence does not require any
One filter that appears in the auxiliary regres- detailed explanation here. However the estima-
sion of the HEGY-test at the complex frequency tion of bR and bI is nonstandard.
but is not present in (1) is Z4,t 5 (L 2 L 3 )Yt . By JS propose the following estimation strategy.
imposing the restriction that this filtered vari- The first step involves regressing D4 Yt , X1t , X2t ,
able has no influence on D4 Yt one assumes the XRt and XIt on lagged values of D4 Yt and Dt , if
absence of so-called non-synchronous seasonal these are present in the model, and defining the
34 ¨ , J. Lyhagen / International Journal of Forecasting 18 (2002) 31 – 44
M. Lof

residuals as R 0t , R 1t , R 2t , R Rt and R It , respective- Since U˜ 0t is a vector and hence vec(U˜ 0t ) 5


ly. If no lags or deterministic variables are U˜ 0t , this yields:
present, Xjt 5 R jt for j 5 1,2,R,I. The restriction
of reduced rank on ai b i9 for i 5 1,2 is removed
by regressing R 0t , R Rt and R It on R 1t and R 2t . It
U˜ 0t 5 U2t F vec( b 9R )
vec( b 9I ) G
1 U˜ ´ t (6)

is shown in JS that the resulting residuals, where 9 ^ a NR ) 2


U2t 5 f (U Rt
defined as U0t ,URt and UIt , satisfy the following N N
(U 9It ^ a I ) (U It9 ^ a R ) 1 (U Rt N
9 ^ a I ) g. The
equation: parameters in bR and bI can now be found from:
U0t 5 2(aR b R9 1 aI b I9 )URt
F G SO DO
T 21 T
vec( b 9R )
5 9
U2tU 2t 9.
U2t U˜ 0t (7)
1 2(aR b I9 2 aI b 9R )UIt 1 U´ t vec( b I9 ) t 51 t51

5 2saR 2 aId SbbR

I
2 bI
bR
9DS D
URt
UIt
1 U´ t For a given value of b, which is generated
randomly in the first iteration, estimates of a˜ 5
˜ 9U1t 1 U´ t ,
5 ab (3)
21
(S01 b ( b 9S 11 b )21 ) / 2 and Vˆ 5 S00 2
21 21
S01 b ( b 9S 11 b ) )b 9S10 are computed. In a sec-
asymptotically. Defining the product moments
ond step U0t is normalized and vectorized as
as Sij 5 (1 /T )S Tt 51UitU 9jt for i, j 5 0,1, we have
described above, yielding a new estimate of b
that for fixed values of b the concentrated
for which we can compute a new likelihood
likelihood function with respect to a˜ and hence
function. This procedure is repeated until a
the variance–covariance matrix V take the
suitable convergence criterion is satisfied.
form, apart from a constant:
Kunst and Franses (1998) argue that deter-
21
2
2] u b 9(S11 2 S10 S 00 S01 )b u ministic seasonal dummy variables, which are
L T
max ( b ) 5 uS00 u]]]]]]]. (4) often included unrestrictedly in (1) to handle the
u b 9S11 b u
deterministic part of seasonality, should be
The minimization of (4) cannot be solved as an confined to the seasonal cointegrating relations
eigenvalue problem as in the zero and biannual instead. This is because unrestricted seasonal
frequency cases since b itself has complex intercepts in the SECM may lead to diverging
structure while the product matrices S11 2 seasonal trends, which is unlikely in most cases.
21
S10 S 00 S01 and S11 do not. JS use a switching We restrict the seasonal dummy variables in a
algorithm proposed by Boswijk (1995) where similar way in the JS specification which leads
the maximum likelihood estimator of b is to the extended variables X˜ 2,t 5 (X2,t , cos p(t)),
calculated iteratively: Isolate bR and bI by using X˜ R,t 5 (XR,t , cos p / 2(t)) and X˜ I,t 5 (XI,t , 2 sin p /
a normalized form (V 21 / 2U0t 5 U˜ 0t ) of U0t , 2(t)). This gives the following augmented b
namely: matrix in (3):
U˜ 0t 5 2V 21 / 2 (aR b R9 1 aI b 9I )URt bR 2 bI
1 2V 21 / 2
(aR b 9I 2 aI b R9 )UI,t 1 V
5 a NR b R9 URt 2 a IN b R9 UIt 1 a RN b I9UIt
21 / 2
U´ t
1 rR
bI
rI
2 rI
bR
rR
2
2 a I b I9URt 1 U˜ ´ t ,
N
(5) where rR and rI are the parameters corre-
and then vectorize (5) by using: sponding to cos p / 2(t) and 2 sin p / 2(t),
respectively.
vec(ABC) 5 (C9 ^ A)vec(B) As in Kunst and Franses (1998) we also
¨ , J. Lyhagen / International Journal of Forecasting 18 (2002) 31 – 44
M. Lof 35

include an unrestricted intercept in the model. theory suggests various long-run relations
Henceforth we will denote the Lee specification among them. One example is the difference of
with unrestricted seasonal intercepts and the output and consumption, in logarithms, which
annual restriction imposed as SCM1, which should be stationary according to the theory.
means seasonal cointegration model of type 1. Another example is the logarithmic difference
The same model with restricted seasonal inter- of output and investments, see Kunst and Neus-
cepts, as proposed by Kunst and Franses (1998), ser (1990). However, it is still an open question
will be denoted SCM2. Finally, SCM3 denote how these series should be related at seasonal
the SECM proposed by JS with restricted frequencies, see Wells (1997) for a discussion of
seasonal intercepts. these issues. The data set for Austria covers the
time period 1964:1 to 1994:4, whereas the time
series for Germany run from 1960:1 to 1988:1
3. Tests for seasonal integration and and for the UK from 1957:1 to 1994:1. Unit
cointegration roots in the univariate time series are tested
using the HEGY-test. The test procedure in-
Gross domestic product (Y), private con- vestigates whether the seasonal difference (1 2
sumption (C), gross fixed investment (I), goods B 4 ), which assumes the presence of four unit
exports (X) and real wages (W) are transformed roots, is the appropriate filter compared with
into natural logarithms. The real interest rates other nested filters.
(R) are given in percentage points. The three Results of the tests for seasonal and non-
data sets were previously used in Kunst and seasonal unit roots appear in Table 1. All
Franses (1998), and the motivation for using auxiliary regressions include an intercept (I),
these series is that the neoclassical growth seasonal dummies (D) and a deterministic trend

Table 1
Seasonal integration tests
Var Austria Germany UK
t pˆ 1 t pˆ 2 Fpˆ 3,4 t pˆ 1 t pˆ 2 Fpˆ 3,4 t pˆ 1 t pˆ 2 Fpˆ 3,4
Y 21.78 23.47* 1.81 21.52 22.83* 3.29 23.21 21.38 4.43
C 21.28 21.83 3.14 21.21 21.45 3.41 23.49* 21.85 3.73
I 22.29 24.96* 9.79* 22.83 22.68 6.21 22.68 22.53 13.3*
X 21.49 25.18* 48.6* 20.96 25.13* 39.7* 21.12 22.12 6.82*
W 21.69 23.31* 2.58 21.18 21.62 2.18 22.16 23.05* 19.9*
R 23.01* 22.95* 7.20* 22.77 22.03 4.73 21.94 25.38* 22.3*

Lags of dependent variable and additional regressors (F ).


Lags F Lags F Lags F
Y 1–4 I,D,T 1 I,D,T 1–5 I,D,T
C 1–4 I,D,T 1–4 I,D,T 1–5 I,D,T
I 1–2 I,D,T 1–3 I,D,T 1 I,D,T
X – I,D,T – I,D,T 1–4 I,D,T
W 1–6 I,D,T 1–4 I,D,T 1 I,D,T
R 1–3 I,D 1 I,D 1–5 I,D
*Significant at the 5% level. I, D and T denote an intercept, seasonal dummies and a trend, respectively.
36 ¨ , J. Lyhagen / International Journal of Forecasting 18 (2002) 31 – 44
M. Lof

(T), except those for the real interest rates presented above for Austria, Finland, Germany
where an exclusion of the trend seems to be a and the UK, mention that evidence of stochastic
more appropriate specification. All variables seasonality is quite weak in the UK series, while
seem to contain unit roots at the zero frequency, the evidence is rather strong in the German data
except UK consumption and the real interest set. Finally, Kunst and Neusser (1990) find
rate in Austria. The results are more mixed at similar unit root results as here, for Austria.
the seasonal frequencies. All roots at the bian- The ranks of the matrices ai b 9i (i 5 1,2) and
nual frequency, except for consumption, are ˜ 9 are determined using the trace test, where
ab
rejected for Austria. On the other hand we find the null hypothesis is: at most r cointegrating
evidence of four unit roots at this frequency for vectors against the alternative of full rank. Table
both Germany and UK. Turning to the annual 2 summarizes the result using the three different
frequency we find evidence of three, one and model specifications SCM1, SCM2 and SCM3.
four stationary variables for Austria, Germany We use the same lag lengths as Kunst and
and UK, respectively. Since unit roots are most Franses (1998) for SCM1 and SCM2, indicated
frequent in the German data set it is sensible to by p, and we use the same lag lengths for SCM3
conclude that cointegration at all frequencies is as well. A* indicates significance at the 5%
most likely to be found there. level. Critical values for columns two, three and
Kunst (1993b), who analyze the variables five are based on our own calculations with

Table 2
Tests for seasonal cointegration, LR-test statistics
Rank Frequency
0 p p/2
SCM1,2,3 SCM1 SCM2,3 SCM1 SCM2 SCM3
Austria, p50
0 161.3* 181.0* 161.4* 200.1* 187.9* 284.7*
1 88.1* 116.5* 113.1* 122.8* 111.1* 173.7*
2 48.8* 77.2* 73.8* 77.8* 67.2* 106.0*
3 22.5 46.1* 42.7* 40.1* 41.7 64.5*
4 6.8 21.0* 20.9* 18.9 21.3 29.7
5 1.5 9.7* 9.7* 6.2 7.8 8.3
Germany, p50
0 132.2* 140.8* 128.9* 174.5* 171.2* 287.3*
1 68.8* 94.1* 82.7* 95.1* 93.6* 179.2*
2 38.2 52.3* 40.8 54.5 56.8 107.6*
3 17.2 24.3 16.8 29.8 32.0 56.9
4 8.4 7.9 8.6 15.2 17.4 27.5
5 1.9 3.2 4.1 4.0 5.9 10.6
UK, p55
0 140.7* 104.0* 95.9 130.7* 130.2 218.7*
1 73.6* 64.5 58.2 71.7 71.5 143.4*
2 47.7* 42.0 39.4 40.6 40.4 92.0
3 25.7 25.3 22.7 22.4 22.1 56.3
4 8.7 10.5 11.3 7.7 7.6 27.8
5 0.8 2.8 3.2 1.7 1.6 6.5
*, Significant at the 5% level. p denotes the lag length.
¨ , J. Lyhagen / International Journal of Forecasting 18 (2002) 31 – 44
M. Lof 37

100 000 replicates and with T 5 400. Lee and ing to the Schwarz Criterion, in addition to
Siklos (1995) present critical values for these equation by equation diagnostic tests. The em-
cases, but they only consider smaller systems. pirical data set for Germany, which covers the
Critical values for columns four and six are time period 1960:1–1988:1 is used. We impose
from Table 1a–f in Franses and Kunst (1999). the ranks according to the results found in
For the last column we use the critical values Section 3 for the different models. All forecast
from Table 2 in JS. Notice that the results are errors correspond to level variables. In the first
identical for the three model specifications at step we save eight observations at the end to be
the zero frequency and for SCM2 and SCM3 at forecasted ex ante. In the next step, the estima-
the biannual frequency. Three long-run cointeg- tion period is extended by one quarter and we
rating vectors are found at the zero frequency reestimate the parameters in the models. How-
for Austria and the UK and two for Germany. ever, we do not change lag orders or the chosen
On frequency p we find no evidence of unit cointegration ranks. These extensions of the
roots for Austria but identify two cointegrating sample are through 1987:4, where we generate
vectors for Germany. For the UK we find no the last one-step ahead forecast error. Hence
evidence of cointegrating vectors at frequency p there are eight one-step ahead forecast errors,
using SCM2 or SCM3, but we identify one if seven two-step ahead forecast errors and so on.
SCM1 is used. The results are more mixed at The eight-step ahead forecasts are then based on
frequency p / 2. One observation is that SCM3 a single observation for each equation. In total
suggests more cointegrating vectors. Based on we have 36 forecasts for each model and data
these results, where the biannual frequency for set. The results are summarized by the RMSE
Austria has full rank and the rather weak statistic (root mean squared error) for 1, 2 and
evidence of cointegration at the seasonal fre- 4-step ahead forecasts. In Table 3 below we also
quencies in the UK data set, only the German consider the RMSE based on all the 36 forecasts
data set is used in Section 4. from each model, denoted All. Finally, we
Having chosen the rank for SCM3 at the present the average performance rank of the
different frequencies we test for further reduc- four model specifications at the different fore-
tion of the model. A hypothesis that simplifies cast horizons considered, denoted AR.
the cointegration analysis on frequency p / 2 is Comparing the four cointegration specifica-
the one implying real structure (HREAL ), i.e. tions we see that the result is quite clear, when
bI 5 0 in (2). However this hypothesis is strong- looking at the average rank of the models. The
ly rejected with LR-test statistics 43.5, 72.5 and version of the seasonal cointegration model with
22.4 for Austria, Germany and the UK respec- the annual restriction and unrestricted seasonal
tively. dummies included (SCM1) is better than the
other three specifications if one step ahead
forecasts are considered. However, for longer
4. An empirical forecasting example forecast horizons the two versions of the SECM
with restricted seasonal dummies seems to be a
In this section we investigate the forecasting better choice. This result is in line with that
performance of the models presented in previ- found in Kunst and Franses (1998). Comparing
ous sections, namely SCM1, SCM2, SCM3. We SCM2 and SCM3 one can see that SCM2 is
also include a VECM in first differences, de- clearly better for forecast horizons shorter than
noted CM. For the seasonal cointegration four steps. The model in first differences (CM)
models we use the same lag lengths as in generates better forecasts than the seasonal
Section 3 and we choose p 5 3 for CM accord- cointegration models at all horizons for goods
38 ¨ , J. Lyhagen / International Journal of Forecasting 18 (2002) 31 – 44
M. Lof

Table 3
Forecast performance. The ranks at different frequencies according to trace test result in Table 2 for the German data set a
Horizon: 1 2 4 All Horizon: 1 2 4 All
SCM1 SCM2
Y 1.09 1.05 1.42 1.16 Y 1.33 0.97 1.29 1.24
] ] ] ]
C 0.84 0.99 0.90 0.84 C 0.91 0.80 0.84 0.77
] ] ] ]
I 3.94 2.87 3.62 3.57 I 4.07 3.03 3.64 3.73
] ] ]
X 2.56 3.17 4.76 4.04 X 2.65 3.02 4.61 3.97
W 1.49 1.70 1.93 2.18 W 1.61 1.58 1.79 2.14
] ]
R 1.35 1.54 1.84 1.52 R 1.40 1.40 1.72 1.45
] ] ]
AR 1.33 2.17 3.00 2.17 AR 2.67 1.50 2.00 2.00
] ] ] ]
SCM3 CM
Y 1.30 1.20 1.39 1.27 Y 1.30 1.75 1.79 1.60
C 1.23 1.12 0.86 0.98 C 1.19 1.47 1.29 1.27
I 4.28 3.26 3.37 3.58 I 4.74 4.55 3.77 4.08
]
X 2.29 2.53 4.35 3.73 X 1.85 2.46 4.14 3.55
] ] ] ]
W 1.61 1.77 1.74 2.09 W 1.86 2.25 2.09 2.48
] ]
R 1.94 1.93 1.89 1.88 R 2.39 2.55 1.49 2.03
]
AR 2.83 2.83 2.00 2.33 AR 3.17 3.50 3.00 3.50
]
a
Forecast errors evaluated with RMSE3100. Unrestricted constant and seasonal intercepts included. All models includes
two vectors at the zero frequency. SCM2 and SCM3 includes two cointegration vectors at the biannual frequency, while
SCM1 include three. Models SCM1 and SCM2 includes two vectors at the annual frequency, while SCM3 include three. AR
denote the average rank of the models, at different forecast horizons, and the smallest is underlined.

exports (X). Note that the result of the HEGY- i.e. no deterministic variables are included. The
tests indicate no seasonal unit roots for goods systems include three variables in each case.
exports in this data set, which may explain why Four different DGPs are investigated, see Table
the model in first differences generates the best 4, and they are denoted DGP1, DGP2, DGP3
forecasts for this time series. However, CM and DGP4. The matrices a1 and b1 concern the
generates the poorest forecasts on average. If zero frequency and include the adjustment
the forecast accuracies are evaluated using the parameters and the cointegrating vectors, re-
mean absolute error (MAE) instead, we come to spectively. The matrices a2 and b2 concerns the
the same conclusions (not shown here). biannual frequency, whereas aR , bR and aI , bI
concerns the annual frequency. The cointegrat-
ing vectors and adjustment parameters are dif-
5. The Monte Carlo setup ferent across frequencies for all DGPs and the
adjustment parameters are always different ac-
The seasonal cointegration specifications we ross DGPs for the annual frequency. We let the
compare are those of Lee (SCM1) and JS columns in all the adjustment parameter ma-
(SCM3), respectively and we set p 5 0 for both trices for a certain DGP sum to the same
specifications. We compare the forecast accura- absolute value, just to have a similar parameter
cy for the seasonal cointegration models when structure across frequencies. For example the
seasonality is viewed as being purely stochastic, first column sum is 0.9 in absolute value across
¨ , J. Lyhagen / International Journal of Forecasting 18 (2002) 31 – 44
M. Lof 39

Table 4
Cointegrating vectors ( b ) and adjustment parameters (a ) used in the Monte Carlo study
DGP1 DGP2 DGP3 DGP4
1.00 0.00 21.00 0.00 1.00 0.00 1.00 0.00
b1 21.00 1.00 21.00 1.00 20.80 1.00 20.80 1.00
0.00 0.50 0.00 0.50 0.00 0.30 0.00 2.30
1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.00
b2 20.90 1.00 20.90 1.00 20.70 1.00 20.70 1.00
0.00 0.30 0.00 0.30 0.00 0.60 0.00 0.60
1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.00
bR 20.70 1.00 20.70 1.00 20.90 1.00 20.90 1.00
0.00 0.60 0.00 0.60 0.00 0.30 0.00 0.30
1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.00
bI 20.80 1.00 20.80 1.00 21.00 1.00 21.00 1.00
0.00 0.30 0.00 0.30 0.00 0.50 0.00 0.50
20.80 0.00 20.40 0.00 20.40 0.00 20.80 0.00
a1 20.10 20.30 20.05 20.15 20.20 20.20 20.10 20.30
0.00 20.20 0.00 20.10 0.00 20.10 0.00 20.20
20.60 0.00 20.30 0.00 20.35 0.00 20.60 0.00
a2 20.30 20.40 20.15 20.20 20.25 20.15 20.30 20.40
0.00 20.10 0.00 20.05 0.00 20.15 0.00 20.10
0.10 0.00 0.05 0.00 0.50 0.00 0.80 0.00
aR 0.80 0.10 0.40 0.05 0.10 0.10 0.10 0.40
0.00 0.40 0.00 0.20 0.00 0.20 0.00 0.10
0.80 0.00 0.40 0.00 0.40 0.00 0.85 0.00
aI 0.10 0.40 0.05 0.20 0.20 0.25 0.05 0.42
0.00 0.10 0.00 0.05 0.00 0.05 0.00 0.08

frequencies in DGP1, whereas the second col- Table 5


umn sum is 0.5. The adjustment parameters are Diagonal and non-diagonal variance–covariance matrices
used in the Monte Carlo study An identity matrix is also
lower in DGP2 than in DGP1. In DGP3 the included i the analysis, denoted S3
values in aR and aI lead to lower parameter
values in the PI 5 (aR b 9I 2 aI b R9 ) matrix, so
we may expect a lower impact of non-syn-
chronous cointegration. Finally, in DGP4 the
S1 5 F 0.7 0.0 0.0
0.0 0.8 0.0
0.0 0.0 0.5
G FS2 5
0.7
0.3
20.2
0.3
0.8
20.1
20.2
20.1
0.5
G
filter XIt should play an even smaller role, when
forecasting.
We investigate the following true rank cases: are used in each case (see Table 5) and four
different simple sizes, namely T 5 40,80,120
1: r 0 5 1, r p 5 1, r p / 2 5 1,
and 200. Furthermore, 100 presample observa-
2: r 0 5 2, r p 5 2, r p / 2 5 1,
tions and 12 postsample observations are gener-
3: r 0 5 1, r p 5 1, r p / 2 5 2, ated. These last 12 observations are saved and
4: r 0 5 2, r p 5 2, r p / 2 5 2. used for ex-ante forecasting. In each case 5000
Three different variance–covariance matrices replications (N) are generated.
40 ¨ , J. Lyhagen / International Journal of Forecasting 18 (2002) 31 – 44
M. Lof

The data is generated in the following way: graphically when DGP1 is used, because SCM3
gives the best forecasts in this case, regardless
Xt 5 Xt24 1 P1 X1,t 1 P2 X2,t of which sample size or true rank is used.
1 PR XR,t 1 PI XI,t 1 ´t , Moreover, the true rank cases 2 and 3 produce
similar results as compared to cases 1 and 4,
where ´t | N3 (0,S1,2,3 ) and where the different respectively, so we restrict the presentation to
P matrices are changed according to which the latter two cases (rank51 and 2 at all
rank is chosen at a particular frequency. For frequencies). Furthermore, if the sample size is
example P1 equals the first r 0 columns in a1 larger than or equal to 80 SCM3 always gener-
times the first r 0 rows in b 19 . The matrices at the ates lower forecast errors, at all horizons, than
biannual and the annual frequency are con- SCM1, if DGP2 is used. As the result for DGP1
structed in the same manner, but the matrices the difference between the two specifications is
PR and PI at the annual frequency equal sometimes substantial, when using DGP2.
(aR b 9R 1 aI b I9 ) and (aR b I9 2 aI b 9R ), respective- Note that there seems to be a seasonal pattern
ly. in the forecast results, see Figs. 1 and 2. This is
For both models we calculate squared fore- due to the estimated variances used in (8) but
cast errors (e 2jk ) for each horizon, k 5 1, . . . ,12, also because the results are for data in levels,
and for each equation, j 5 1, 2 and 3. Then we hence transformed from fourth differences.
take the equation by equation average of these SCM3 produces poorer forecasts for smaller
over the number of replications (R). We then sample sizes if DGP3 and DGP4 are considered,
weigh the resulting mean squared forecast errors see for example (c), (d), (e) and (f) in Fig. 1.
with a variance estimate of the 12 first observa- This is because the values in aR and aI result in
tions generated from each equation, denoted Vˆjk . a lower impact of so-called non-synchronous
The variances are calculated after 5000 replica- seasonal cycles. Some results indicate that
tions and they are useful when we, in the last SCM3 may yield worse forecasts if more coin-
step, average the mean squared forecast errors tegrating relations are included in the model.
over the three equations. With this relative One example is (a) and (b) in Fig. 1, for DGP2
measure we avoid to give high weight to and T 5 40. Looking at the rest of the figures
variables with large variances. To summarize, increasing sample size works in favor of SCM3,
for each model and horizon we have the follow- so that for T 5 200 SCM3 dominate SCM1. We
ing mean squared error [MSE] measure: also find in a smaller complimentary study (not
all 192 combinations considered) that these
O FS O DYVˆ G.
3 R
1 1 results hold when various lag lengths are select-
MSE 5 ] ] e 2jk (8)
3 j 51 R r51 jk
ed, using the Akaike Information Criterion.

6. Monte Carlo results


7. Conclusions
In total we have 192 combinations to consi-
der if we look at all sample sizes, true rank The forecasting performance of the seasonal-
cases, DGPs and variance–covariance matrices. ly cointegrated model of Johansen and Schaum-
However, the results are not different depending burg (1998) is compared to a related specifica-
on which variance–covariance matrix is used. tion, with a restriction at the annual frequency.
Because of that only results based on S1 are In the empirical forecasting part, we also in-
considered. We choose not to report any results clude a model in first differences with co-
¨ , J. Lyhagen / International Journal of Forecasting 18 (2002) 31 – 44
M. Lof 41

Fig. 1. Mean squared errors weighted with variances. Dashed line: SCM1, solid line: SCM3. Rank concerns all frequencies,
T540 and 80.
42 ¨ , J. Lyhagen / International Journal of Forecasting 18 (2002) 31 – 44
M. Lof

Fig. 2. Mean squared errors weighted with variances. Dashed line: SCM1, solid line: SCM3. Rank concerns all frequencies,
T5120 and 200.
¨ , J. Lyhagen / International Journal of Forecasting 18 (2002) 31 – 44
M. Lof 43

integration restrictions. We examine data sets This sample size effect may explain the result in
from Austria, Germany and the UK, each the empirical example which is based on fewer
containing six variables: gross domestic prod- observations.
uct, private consumption, gross fixed invest-
ment, goods exports, real wages and the real
interest rate. We examine the integration and Acknowledgements
cointegration properties for the three data sets ¨
We thank Lars-Erik Oller, the Editor and
and consider the effect of including restricted or
three anonymous referees for helpful comments.
unrestricted seasonal dummies in the seasonal
We also thank Robert Kunst who generously
cointegration models. The biannual frequency
allowed us to use the data sets. Financial
for Austria seems to have full rank and we find
support for Lyhagen from Tore Browaldh’s
rather weak evidence of cointegration at the
Foundation for Scientific Research and Higher
seasonal frequencies in the UK data set, so that Education and The Swedish Foundation for
only the German data set is used in the empiri- International Cooperation in Research and
cal forecasting example. Higher Education (STINT) is gratefully ack-
The seasonal cointegration model with the nowledged. Lyhagen is also grateful to Prof. M.
annual restriction and unrestricted seasonal H. Pesaran who acted as his sponsor at a Post-
dummies included is better than the other three Doc stay with the Faculty of Economics and
specifications if one step ahead forecasts are Politics, University of Cambridge.
considered. However, for longer forecast
horizons the two versions of the seasonal coin-
tegration model with restricted seasonal dum- References
mies seems to be a better choice. When compar-
ing the two restricted seasonal dummy spe- Boswijk, H. P. (1995). Identifiability of cointegrated
cifications we find that the version with the systems. Discussion paper, Tinbergen Institute, TI 95,
78. (http: / / www.fee.uva.nl / ke / boswijk / main.htm)
annual restriction is better for forecast horizons Franses, P. H., & Kunst, R. M. (1999). On the role of
shorter than four steps. The cointegration model seasonal intercepts in seasonal cointegration. Oxford
in first differences generates the poorest fore- Bulletin of Economics and Statistics 61, 409–434.
casts on average. Hylleberg, S., Engle, R. F., Granger, C. W. J., & Yoo, B. S.
In the Monte Carlo study we analyze four (1990). Seasonal integration and cointegration. Journal
of Econometrics 44, 215–238.
different DGPs and four different sample sizes, Johansen, S. (1988). Statistical analysis of cointegration
namely T 5 40, 80, 120 and 200 when using a vectors. Journal of Economic Dynamics and Control
three variate system. We do not include any 12, 231–254.
deterministic variables and seasonality is thus Johansen, S. (1995). Likelihood-based inference in co-
generated as being purely stochastic and we do integrated vector autoregressive models, Oxford Uni-
versity Press, Oxford.
not consider any model in first differences. Johansen, S., & Schaumburg, E. (1998). Likelihood analy-
When the smaller sample sizes are considered, sis of seasonal cointegration. Journal of Econometrics
we find some evidence that the specification 88, 301–339.
proposed by Johansen and Schaumburg (1998) Kunst, R. M. (1993a). Seasonal cointegration, common
may yield worse forecasts if more cointegrating seasonals, and forecasting seasonal series. Empirical
Economics 18, 761–776.
relations are included in the model. Increasing
Kunst, R. M. (1993b). Seasonal cointegration in macro-
sample size works in favor of this model, so that economic systems: case studies for small and large
for T 5 200 it dominate the seasonal cointegra- European countries. Review of Economics and Statistics
tion model with the annual restriction included. LXXV, 325–330.
44 ¨ , J. Lyhagen / International Journal of Forecasting 18 (2002) 31 – 44
M. Lof

Kunst, R. M., & Franses, P. H. (1998). The impact of Wells, J. M. (1997). Business cycles, seasonal cycles and
seasonal constants oncasting seasonally cointegrated common trends. Journal of Macroeconomics 19, 443–
time series. Journal of Forecasting 17, 109–124. 469.
Kunst, R. M., & Neusser, K. (1990). Cointegration in a
macroeconomic system. Journal of Applied Economet- ˚
Biographies: Marten LOF¨ is a Ph.D. student at the
rics 5, 351–365. Department of Economic Statistics, Stockholm School of
Lee, H. S. (1992). Maximum likelihood inference on Economics. His research focuses on seasonality, cointegra-
cointegration and seasonal cointegration. Journal of tion and forecasting.
Econometrics 54, 1–47.
Lee, H. S., & Siklos, P. (1995). A note on the critical Dr. Johan LYHAGEN obtained his Ph.D. in Statistics from
values for the maximum likelihood (seasonal) cointegra- Uppsala University in 1997. In his thesis he examined
tion tests. Economic Letters 49, 137–145. different aspects of fractional ARMA models. His current
Reimers, H. -E. (1997). Forecasting of seasonal cointe- research interests are time series analysis, cointegration
grated processes. International Journal of Forecasting and the analysis of financial data.
13, 369–380.

You might also like