This short paper is a straightforward continuation of Sauramo (1998). The aim of that paper was to analyse the boom of the late eighties and the depression of the early nineties by estimating structural VAR models which were based on the utilization of the traditional aggregate demand6aggregate supply framework. These models typically belong to the class of IS-LM models which have been augmented with a Phillips curve.

2. THE BOOM AND THE DEPRESSION:
A NOTE ON THE IDENTIFICATION OF
AGGREGATE SUPPLY SHOCKS
Pekka Sauramo*
Labour Institute for Ecnomic Research
* Most of this paper was written during my stay at New York University as a visiting
scholar. I would like to thank Jordi Gali for the opportunity to utilize some of his
software while conducting the analysis. The usual disclaimer applies. Financial
support from the Yrjö Jahnsson Foundation is gratefully acknowledged.
LABOUR INSTITUTE FOR ECONOMIC RESEARCH
DISCUSSION PAPERS 147
HELSINKI 1998

3. ISBN 952$$5071$$25$$1
ISSN 1236$$7184

4. 3
TIIVISTELMÄ
Tämän menetelmäpainotteisen tutkimusselosteen analyysilla on kaksi
pääasiallista tavoitetta: 1) muodostaa kokonaiskysyntä-kokonaistarjonta
-kehikon puitteissa muuttuja, joka kuvaisi teknologista edistystä ja 2)
analysoida teknologisen kehityksen merkitystä kokonaistuotannon vaihte-
luiden selittäjänä 1980-luvun nousukaudella ja 1990-luvun alun lamavuosi-
na.
Suomalaisen laman syitä tarkastelevan keskustelun kannalta analyysi
muodostaa vain yhden sivujuonteen. Eihän keskustelussa mikään keskus-
teluosapuoli ole korostanut teknologiashokkien merkitystä. Kun laman syitä
aletaan tarkastella ekonometrisesti kokonaiskysyntä-kokonaistarjonta -ke-
hikossa, teknologiashokkien identifiointi osana muiden $ ja mahdollisesti
tärkeämpien $ shokkien identifiointia on kuitenkin analyysin oleellinen osa.
Tässä tutkimusselosteessa teknologiashokit identifioidaan osana raken-
teellisen VAR mallin identifiointia olettamalla, että vain ne voivat muuttaa
työn tuottavuuden tasoa pysyvästi. Muut shokit $ esimerkiksi kokonais-
kysyntäshokit $ voivat muuttaa työn tuottavuuden tasoa vain tilapäisesti.
Oletus voidaan liittää osaksi esimerkiksi uuskeynesiläistä dynaamista
yleisen tasapainon mallia.
Vaikka kokonaiskysyntäshokit eivät tutkimusselosteen mallissa voikaan
muuttaa työn tuottavuuden tasoa pysyvästi, ne voivat muuttaa pysyvästi
tuotannon tasoa. Siten mallissa luovutaan oletuksesta vertikaalisesta
pitkän aikavälin Phillipsin käyrästä.
Tutkimusselosteessa käytetty oletus teknologiashokkien muodostamiseksi
tuotti järkeviä tuloksia. Positiiviset teknologiashokit olivat deflatorisia ja työn
kysyntää ainakin lyhyellä aikavälillä vähentäviä. Niiden merkitys 1980-
luvun lopun nousukauden ja 1990-luvun alun lamavuosien talouskehityk-
sen muovaajana oli kuitenkin $ odotetusti $ vähäinen. Näinä vuosina
kokonaiskysyntäshokkien merkitys oli ratkaiseva.

5. 4
1. INTRODUCTION
This short paper is a straightforward continuation of Sauramo (1998). The
aim of that paper was to analyse the boom of the late eighties and the
depression of the early nineties by estimating structural VAR models which
were based on the utilization of the traditional aggregate
demand$aggregate supply framework. These models typically belong to
the class of IS-LM models which have been augmented with a Phillips
curve.
A large number of recent econometric studies have had this framework as
the point of departure. It provides one way of linking econometric
investigation about economic fluctuations to economic theory. One can
therefore expect that the framework is superior to largely atheoretic
frameworks such as the ones utilized, for example, in Blanchard (1993) or
Sauramo (1996), because the interpretation of shocks should become
easier.
However, the usefulness of that kind of investigation crucially depends on
how well the framework fits the data. Finding a suitable framework for
Finland is not easy. Until the mid eighties, the financial markets were
regulated and therefore the standard textbook versions were more or less
inapplicaple in the description of the behaviour of the economy. The
deregulation of the financial markets has brought about a drastic change,
but this does not necessarily make the analysis easier: institutional
changes were accompanied by shifts in the exchange rate policy and
monetary policy regimes, which complicates the use of the standard IS-LM
framework or its cousin, the Mundell-Fleming framework.
The paper illustrated that, when the Finnish data is used, the identification
of shocks which would be compatible with the conventional aggregate
demand $ aggregate supply framework was difficult. In particular, the
identification of aggregate supply shocks turned out to be a very knotty
task.

6. 5
Within the traditional aggregate demand$aggregate supply framework
positive aggregate supply shocks (for example, technology shocks or
labour supply shocks) are deflationary, i.e. they decrease prices. The
shocks which were supposed to be (positive) aggregate supply shocks
turned out to be inflationary, however. Thus the shocks did not satisfy this
over-identifying restriction.
In the paper, as in numerous other papers, the classification of shocks into
aggregate demand and aggregate supply shocks rests on the assumption
about the vertical long-run supply (or Phillips) curve. If that is assumed to
exist, aggregate demand shocks do not affect output in the long run.
Positive aggregate demand shocks, which are inflationary, can have only
a transitory influence on real output. In the long run they only increase
prices. Consequently, aggregate demand and aggregate supply shocks
can be identified by using a long-run (zero) restriction. The use of such a
restriction was pioneered by Blanchard and Quah (1989).
The assumption about the vertical long-run Phillips curve is, of course,
problematic. It is easy to think of channels through which aggregate
demand shocks may have a long-lasting, or even permanent, effect on
output (capital accumulation, hysteresis in the labour market, increasing
returns to scale etc.) One of the main conclusions drawn in Sauramo
(1998) was that the assumption about the vertical long-run Phllips curve
may not be well-grounded in the case of Finland.
This conclusion is consistent with the view that, traditionally, economic
growth has been demand-led in Finland with demand for exports being the
main source of the growth. However, economic developments during the
past ten years have been so peculiar in Finland that the results of Sauramo
(1998) should not be interpreted as simply reflecting this traditional Finnish
growth pattern.
Even though the framework based on the vertical long-run Phillips curve
does not seem to work very well in the case of Finland, it does not mean
that there does not exist an aggregate demand$aggregate supply

7. 6
framework which would be useful in the description of economic
fluctuations in Finland.
In Sauramo (1998) the main difficulty was the identification of aggregate
supply shocks. The assumption about the vertical long-run Phillips curve
did not produce shocks which could be interpreted as, for example,
technology or labour supply shocks. In this paper I attempt to identify
aggregate supply shocks (i.e. technology shocks) by utilizing identifying
assumptions which do not necessarily imply the existence of a long-run
vertical Phillips curve.
The basic idea is very simple: in using a structural VAR model I identify
technology shocks by assuming that they are the only source of permanent
changes in the level of labour productivity. Other shocks, whether they are
aggregate demand or supply shocks, can have a permanent influence on
the level of output. In this sense the assumption used in this paper is more
general than the one used originally in Blanchard and Quah (1989). This
paper draws mainly on Gali (1996) in which the main identifying restriction
is rationalized by a new Keynesian dynamic general equlibrium model.
The main finding of this paper is that the identifying long-run restriction
yields plausible technology shocks. They are plausible in the sense that
positive technology shocks are deflationary and, as the theoretical model
suggests, reduce demand for labour at least in the short run. However,
they did not play a major role either in the boom or in the depression. The
boom and the depression are explained by non-technology shocks. As in
Sauramo (1998), the best way of characterizing these shocks is to regard
them as aggregate demand shocks.
In the next chapter, I briefly discuss the framework and the data. Chapter 3
contains the main results and Chapter 4 concludes the work.

8. 7
2. THE FRAMEWORK AND THE DATA
As noted in the introduction, the main difficulty in Sauramo (1998) was the
identification of aggregate supply shocks. The assumption about the
vertical long-run Phillips curve did not produce shocks which could be
interpreted as, for example, technology or labour supply shocks. One
should be able, in principle, to identify aggregate demand and aggregate
supply shocks by utilizing a simple bivariate model and using the identifying
restriction that only supply shocks have permanent effects on output.
Blanchard and Quah (1989), in their pioneering work, employed a model
which describes the joint behaviour of GNP and the unemployment rate.
This choice is, of course, not the only alternative.
In Sauramo (1998) I estimated a bivariate model by using data on GDP and
consumer prices. Within that model positive aggregate demand shocks
should raise consumer prices both in the short and in the long run, while
positive supply shocks should reduce prices. Because these responses are
not imposed, they can serve as over-identifying restrictions when the sense
of the results is discussed. The main conclusion was that the shocks which
were identified did not satisfy these over-identifying restrictions. Shocks
which were supposed to be positive supply shocks were inflationary.
I also estimated three-, four-, and five-dimensional models, with interest
rates, private consumption and the terms of trade as additional variables.
In not a single experiment did the assumption about the vertical long-run
Phillips curve produce plausible supply shocks. They were like demand
shocks which have a permanent influence on output $ a feature which
cannot be reconciled with the assumption about the vertical long-run
Phillips curve.
The Finnish evidence therefore seems to suggest that one should be ready
to rely on frameworks which are not based on the vertical long-run Phillips
curve paradigm. Fortunately, it is very easy to go beyond this paradigm by
still retaining the aggregate demand$aggregate supply framework.

9. 1
In this short note I will not survey the relevant literature thoroughly. For a
discussion, see Gali (1996).
8
In this paper the assumption about the long-run Phillps curve is replaced by
an assumption which allows one to attempt to identify plausible technology
shocks. There are also, of course, other supply shocks than technology
shocks, for example, shocks to the labour supply. However, the main
purpose of this paper is not to try to identify all potentially relevant
aggregate supply shocks but rather to analyse whether it is possible, by
using Finnish data, to identify at least one class of relevant aggregate
supply shocks.
In this paper, the identification of technology shocks is based on the
following identifying assumption: only technology shocks have a permanent
effect on the level of labour productivity. A broad class of theoretical
models satisfies this restriction. It includes, among others, real business
cycle (RBC) models and new Keynesian models. This paper rests on Gali
(1996), which utilizes a new Keynesian general equilibrium model.1
Essentially the same identifying asssumption has also been used, for
example, by Dolado and Jimeno (1997). (See also Castillo, Dolado and
Jimeno 1998, Jacobson, Vredin, and Warne 1997, 1998.)
The identification of technology shocks can be based on the use of simple
bivariate models which utilize data on real output and labour input. For
example, Gali (1996) estimates bivariate models with labour productivity
and labour input being the two variables.
In this paper I tie the analysis as closely as possible to that of Sauramo
(1998). Therefore, in addition to data on output and labour input I also
utilize data on consumer prices. It allows one to check whether the relevant
shocks are deflationary or inflationary. Positive technology shocks should
be deflationary. Because this kind of response is not imposed, it can
serve as an over-identifying restriction when the sense of the results is
discussed.

10. 2
This kind of response is clearly in contrast with predictions of the conventional RBC
models. In those models positive technology shocks, by shifting demand for labour
schedules, have a positive effect on the level of employment.
3
The relevant results are those which were obtained when the bivariate output $
prices model was estimated.
9
Within the new Keynesian framework employed by Gali (1996), positive
technology shocks should have a negative short-run effect on the level of
employment. The intuition for this result is obvious, especially in the
special case when the money supply is assumed to be exogenous. In that
case a constant money supply and predetermined prices imply that real
balances and, consequently, aggregate demand, and output, remain
unchanged during the period when the technology shock occurs. If the
technology shock is positive, the same output can be produced by less
labour input. 2
This kind of response can also be used as an additional
over-identifying restriction.
I use the same quarterly data on GDP and the Consumer Price Index as in
Sauramo (1998). For labour input I use quarterly data on total hours
worked by employees (according to Labour Force Statistics). Alternatively,
one could measure labour input by utilizing data on the number of
employees. For the present purpose, data on hours is likely to be
preferable. The data covers the same period as in Sauramo (1998), i.e.
the years 1975:1$1996:2. The series are seasonally adjusted.
Figures 1 and 2 depict the data both in logs and log differences. According
to the standard Dickey-Fuller tests (log of) output (y) and (log of) labour
productivity (y-h) are I(1) variables. Even though the best way of
characterizing (log of) consumer prices (p) can be to regard it as a I(2)
variable, the analysis will be performed by assuming that p is a I(1)
variable. This assumption allows one to compare the results of this paper
with the relevant results of Sauramo (1998).3

11. 10
Figure 1: Data on GDP, hours, productivity and prices
Seasonally adjusted data in levels: 1975:1-1996:2
log of GDP
75 78 81 84 87 90 93
11.2
11.3
11.4
11.5
11.6
11.7
11.8
log of hours
76 79 82 85 88 91 94
13.40
13.45
13.50
13.55
13.60
13.65
13.70
13.75
log of productivity
75 78 81 84 87 90 93
-2.32
-2.24
-2.16
-2.08
-2.00
-1.92
-1.84
-1.76
-1.68
log of CPI
75 78 81 84 87 90 93
3.4
3.6
3.8
4.0
4.2
4.4
4.6
4.8
Figure 2: Data on GDP, hours, productivity and prices
Seasonally adjusted data in differences: 1975:2-1996:2
difference of log GDP
75 78 81 84 87 90 93
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04
0.05
difference of log hours
75 78 81 84 87 90 93
-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
difference of log productivity
75 78 81 84 87 90 93
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
difference of log CPI
76 79 82 85 88 91 94
-0.008
0.000
0.008
0.016
0.024
0.032
0.040
0.048

12. 4
Thus the model was estimated in the difference form. Three lags were used, with
Schwarz and Hannan-Quin information criteria being the main decision-making
criteria. The estimation period was 1976:1$1996:2.
5
I assume that the reader of this paper has first read Sauramo (1998).
11
3. RESULTS
The results are based on the estimation of a three-variable productivity-
output-prices model. The model is a straightforward extension of the two-
variable output-prices model which served as the benchmark model in
Sauramo (1998). In this paper that model is augmented with a productivity
variable.
For the identification of technology shocks, the use of two variables (for
example, productivity and output, or productivity and hours) would be
enough (see Gali, 1996). However, the use of the three-variable model
enables one to analyse how the shocks, which are supposed to be
technology shocks, affect output and prices. Because positive technology
shocks should be deflationary, the price variable provides one tool for
checking the plausibility of shocks.
The estimation of the unconstrained educed form is based on the
assumption that xt = (

13. yt-

14. ht ,

15. yt ,

16. pt) is a covariance stationary process.4
The identification of shocks takes place in a similar fashion as in Sauramo
(1998) (see pages 10$12).5
The use of three variables enables one to identify three shocks. I use such
identifying constraints that, on the one hand, enable one to interpret one
shock as a technology shock, and, on the other hand, allows one to
compare the results with those obtained when the benchmark model in
Sauramo (1998) is used.

17. 6
Variables are expressed in levels, in the figures, even though the model was
estimated in the difference form. The impulse responses for hours is easy to derive
after computing impulse responses for productivity and output. Confidence bands
for impulse responses were computed by utilizing a Monte Carlo method which is
12
In the three-variable case nine constraints are needed for just-identification.
Six constraints are given by the assumption that shocks are mutually
orthogonal and that their variances equal unity.
Two long-run restrictions are used to identify the technology shocks: it is
assumed that, unlike the technology shock, neither of the other two shocks
affects the level of productivity in the long run.
Finding an uncontroversial additional constraint as the third identifying
constraint which would identify the two other shocks is not straightforward.
Since I want to compare the results of this paper with those of Sauramo
(1998) I use the same identifying restriction that was used in the estimation
of the benchmark model in that paper: it is assumed that one of the three
shocks has only a transitory influence on output.
These restrictions imply that the three shocks have the following
characteristics. The shock which is supposed to be the technology shock is
the only shock which may have a permanent effect on the level of
productivity. The second shock may have a permanent influence on output
and the price level. Therefore, two of the three shocks may affect output
permanently. The third shock can have a permanent effect only on the
price level.
I will not label the other two shocks as aggregate demand or aggregate
supply shocks without analysing their special features. These will be
illustrated by using impulse responses and forecast error variance
decompositions. The importance of the various shocks for economic
fluctuations during the boom and depression years are analysed by utilizing
historical forecast error decompositions.
Figures 3a and 3b display the impulse responses associated with the three
shocks together with one-standard error confidence bands.6
In Figure 3a

18. based on sampling from the estimated asymptotic distribution of the VAR
coefficients and the covariance matrix of the innovations. In each draw the sample
size amounted to 500. I wish to thank Jordi Gali for kindly providing the RATS code
for performing the computations.
13
Figure 3a: Impulse Responses
productivity - GDP-prices -model
Technology Shock
productivity
0 5 10
0.000
0.005
0.010
0.015
0.020
0.025
gdp
0 5 10
-0.0080
-0.0040
0.0000
0.0040
0.0080
0.0120
0.0160
prices
0 5 10
-0.024
-0.016
-0.008
0.000
0.008
hours
0 5 10
-0.024
-0.018
-0.012
-0.006
-0.000
0.006
Demand Shock
productivity
0 5 10
-0.0040
0.0000
0.0040
0.0080
0.0120
gdp
0 5 10
0.010
0.015
0.020
0.025
0.030
prices
0 5 10
-0.01
0.00
0.01
0.02
0.03
hours
0 5 10
0.000
0.005
0.010
0.015
0.020
0.025
0.030

19. 14
Figure 3b:Impulse Responses
productivity - GDP-prices -model
2.non-technology shock
productivity
0 5 10
-0.0030
0.0000
0.0030
0.0060
0.0090
0.0120
0.0150
gdp
0 5 10
-0.0075
-0.0050
-0.0025
0.0000
0.0025
prices
0 5 10
0.0000
0.0045
0.0090
0.0135
0.0180
0.0225
0.0270
hours
0 5 10
-0.0150
-0.0120
-0.0090
-0.0060
-0.0030
-0.0000
0.0030
the left-hand panel depicts the responses to the shock which is supposed
to be a positive technology shock. The impulse responses are consistent

20. 15
with that kind of interpretation. The positive technology shock increases
both the level of productivity and output. Furthermore it decreases hours
and prices at least in the short run. The identification of technology shocks
therefore seems to have been successful.
The right-hand panel of Figure 3a depicts impulse responses which are
associated with the first non-technology shock. The best way of interpreting
the shock is to regard it as a positive aggregate demand shock which is
inflationary and which has a permanent positive effect on the level of
output and employment. It also seems to have a positive effect on the level
of productivity in the short and medium run but (taking into account the
confidence band) this interpretation may be too straightforward. The first
non-technology shock clearly corresponds to the permanent shock in the
benchmark case of Sauramo (1998) (see Figure 2 on page 18). On the
other hand, like the transitory shock in the benchmark case, the transitory
shock in Figure 3b is difficult to interpret (see Figure 3 on page 19). One
possibility is to regard it as a positive transitory cost shock (i.e a transitory
negative supply shock) which is associated with an increase in prices and
a transitory (but long-lasting) decrease in output and employment.
In Sauramo (1998) one of the main conclusions was that fluctuations (and
growth) of GDP seem to have been attributable to permanent aggregate
demand shocks. For comparison, the relative importance of the three
shocks is illustrated in Table 1.

21. 16
Table 1. Forecast-error variance decompositions for GDP:
productivity-output-prices model
Horizon in quarters Technology
shock
Demand shock Transitory
shock
Contemporaneous 5.6 93.7 0.7
4 5.0 93.5 1.5
8 4.3 94.4 1.3
12 3.9 95.1 1.0
16 3.6 95.6 0.8
20 3.3 96.0 0.7
24 3.1 96.3 0.5
100 2.2 97.7 0.1
Note: In the table the forecast error variances have been decomposed to three
sources. For instance, of the 4-step forecast error variance of GDP, 5.0 percent is
accounted for by technology shocks, 93.5 per cent is by demand shocks and 1.5
percent by transitory shocks.
As in Sauramo (1998) (Table 1 on page 21), fluctuations are mainly
attributable to permanent demand shocks. Technology shocks seem to
have been of minor importance. Thus the succesful identification of
technology shocks does not change the main results of the earlier study.
This conclusion is supported by Table 2, which displays the decomposition
of eight-quarter forecast errors for GDP. (Figures are yearly averages
calculated by using quarterly observations. Eight quarters were chosen,
because the peaks and troughs correspond closely enough to the peaks
and the troughs of the GDP series.) Both the boom and the depression
were caused by permanent demand shocks (see also Table 2 in Sauramo
(1998) on page 21).

22. 17
Table 2. Decomposition of eight-quarter forecast errors for GDP:
productivity-output-prices model (yearly averages for 1986$1995)
Year GDP Technology
shock
Demand shock Transitory
shock
1986 0.6 0.4 -0.5 0.7
1987 1.2 0.6 0.4 0.2
1988 4.5 0.1 4.5 -0.1
1989 5.3 0.6 5.0 -0.3
1990 -0.2 0.9 -0.9 -0.2
1991 -13.5 -0.8 -12.6 -0.1
1992 -13.6 -1.4 -11.8 -0.4
1993 -4.5 1.5 -5.1 -0.9
1994 1.6 2.4 -0.4 -0.4
1995 6.1 0.3 5.0 0.8
Note. Forecast errors of the first column are based on the use of the level form of
the model. Consequently, the forecast errors have been obtained by subtracting the
forecast errors from the logs of the realize d levels of GDP.
The figures in Table 2 are interpreted as follows. In 1986 the eight-quarter
forecast error for GDP was 0.6 per cent, i.e. the realized level of GDP was
0.6 per cent higher than the forecast by the model. Owing to the way the
forecast errors are computed, they are not exactly the same as relative
forecast errors. The difference, however, is small. (For the computation of
the forecast errors see the note in Table 2.)
According to Table 2, positive technology shocks did play a noticeable role
during the years 1993 and 1994. They contributed to the recovery of the
economy. During those years positive technology shocks also had a strong

23. 18
deflationary effect. (Historical forecast-error decompositions for prices are
available upon request.)
Even though technology shocks have not played an important role in
explaining fluctuations in GDP they have largely determined developments
of productivity both in the short and long run. This can be seen from Table
3. Also, they are important in explaining short-run fluctuations of hours. For
example, at a four-quarter horizon 35 per cent of the forecast error
variance for hours is accounted for by the technology shocks.
Table 3. Forecast-error variance decompositions for productivity:
productivity-output-prices model
Horizon in quarters Technology
shock
Demand shock Transitory
shock
Contemporaneous 95.1 2.4 2.5
4 93.8 3.1 5.1
8 94.2 2.9 2.9
12 94.6 2.7 2.7
16 95.2 2.4 2.4
20 95.8 2.1 2.1
24 96.2 1.9 1.9
100 99.0 0.5 0.5
Note: See note in Table 1.

24. 19
4. CONCLUSIONS
In Sauramo (1998) I drew the conclusion that in modelling Finnish
economic fluctuations one should be ready to rely on frameworks which are
not based on the vertical long-run Phillips curve paradigm. This conclusion
was based on experimenting with models, most of which belonged to that
paradigm. When such models were used, the main difficulty was the
identification of aggregate supply shocks. Typically, the shocks which were
supposed to be aggregate supply shocks did not satisfy the over-identifying
restriction that positive aggregate supply shocks are deflationary.
In this paper I attempted to identify one class of aggregate supply shocks,
technology shocks, by utilizing a framework and assumptions which do not
necessarily imply the existence of a vertical long-run Phillips curve. Within
the framework of this paper, aggregate demand shocks could have a
permanent influence on output.
Technology shocks were identified by assuming that they are the only
source of permanent changes in the level of labour productivity. The
identification of technology shocks was succesful in the sense that positive
technology shocks were deflationary and they reduced employment in the
short run. Unlike that in RBC models, this kind of response is typical in new
Keynesian general equilibrium models, of which Gali (1996) is the one that
this paper drew on.
Even though the identification of technology shocks was successful it
turned out that their role in shaping economic fluctuations during the boom
years of the late eighties and the depression years of the early nineties was
only minor. This is hardly surprising. However, they did play a noticeable
role during the recovery of the economy, especially in 1994.
In this paper, I have not conducted a detailed analysis of the years of
recovery. Doing that would be a well-motivated next step. The
development of employment should deserve special attention, since those
years have been characterized as the years of “jobless growth”. Obviously,

25. 20
the framework used in this paper is well suited for the examination of the
validity of that kind of characterization.

26. 21
REFERENCES
Blanchard. O. J. (1993), Consumption and the Recession of 1990$1991,
American Economic Review 83, 270$274.
Blanchard, O. J. and D. Quah (1989), The Dynamic Effects of Aggregate
Demand and Supply Disturbances, American Economic Review, 79,
655$673.
Castillo, S., J. J.Dolado and J. F. Jimeno (1998), The tale of two neighbour
economies: labour market dynamics in Spain and Portugal, CEPR
Discussion paper series No. 1954, London.
Dolado, J. J. and J. F. Jimeno (1997), The causes of Spanish
unemployment: A structural VAR approach. European Economic Review,
41, 1281$1307.
Gali, J. (1996), Technology, Employment, and the Business Cycle: Do
Technology Shocks Explain Aggregate Fluctuations, NBER Working Paper
No. 5721.
Jacobson, T., A. Vredin and A. Warne (1997), Common Trends and
Hysteresis in Unemployment. European Economic Review,41, 1781$1816.
Jacobson, T., A. Vredin and A. Warne (1998), Are Real Wages and
Unemployment Related?, Economica, 65, 69$96.
Sauramo, P. (1996), The Boom and the Depression $ A Simple Shock
Interpretation, Discussion papers 132, Labour Institute for Economic
Research, Helsinki.
Sauramo, P. (1998), The Boom and the Depression: An Analysis within the
Aggregate-Demand$Aggregate-Supply Framework, Discussion papers
143, Labour Institute for Economic Research, Helsinki.