SBP Working Paper Series

No. 01 September, 2001

Long-Run Trend, Business Cycles and

Short-Run Shocks in Real GDP

Muhammad Farooq Arby

STATE BANK OF PAKISTAN

 SBP Working Paper Series
Editor: Riaz Riazuddin

The objective of the SBP Working Paper Series is to stimulate and generate discussions, on
different aspects of macroeconomic issues, among the staff members of the State Bank of
Pakistan. Papers published in this series are subject to intense internal review process. The
views expressed in the paper are those of the author(s) and do not necessarily reflect those of
the State Bank of Pakistan.

© State Bank of Pakistan

All rights reserved.

Price per Working Paper

Pakistan: Rs 50 (inclusive of postage)

Foreign: US$ 20 (inclusive of postage)

Purchase orders, accompanied with cheques/drafts drawn in favor of State Bank of Pakistan,
should be sent to:

Chief Spokesperson
Corporate Services Department,
State Bank of Pakistan,
I.I. Chundrigar Road, P.O. Box No. 4456,
Karachi 74000. Pakistan

For all other correspondence:

Editor,
SBP Working Paper Series
Research Department,
State Bank of Pakistan,
I.I. Chundrigar Road, P.O. Box No. 4456,
Karachi 74000. Pakistan

Published by: Editor, SBP Working Paper Series, State Bank of Pakistan, I.I. Chundrigar
Road, Karachi, Pakistan.

http://www.sbp.org.pk

Printed at the SBPBSC (Bank) – Printing Press, Karachi, Pakistan

State Bank of Pakistan Working Paper No. 1/01, September 2001

Long-run Trend, Business Cycles and Short-run Shocks in Real GDP

By
Muhammad Farooq Arby
Assistant Director
General Economic Research Department
State Bank of Pakistan
Central Directorate
Karachi
Email: [email protected] *

Abstract:
Gross domestic product may be considered, a combination of three processes viz., long-run
trend, business cycles and short-run shocks. The series of GDP can be decomposed in to its
three components by using some statistical method. Such a decomposition of real GDP of
Pakistan reveals that the Pakistan’s economy has a declining growth in long-run trend since
early 1980s that however, is expected to start rising in 2001-02. Pakistan is also facing a
recessionary phase of third business cycle, which is expected to end in 2004-05.

Views expressed in this Working Paper are the author’s personal views and do not necessarily reflect those of the
State Bank of Pakistan.
* The author of this paper welcomes comments & suggestions.

2
T he path of economic growth for any country depends on a number of factors including
structural changes in the economy, natural calamities, political instabilities, global
recessionary trends, self-feeding business cycles, etc., and the combined effect of all these
factors is most commonly represented in the country’s Gross Domestic Product. In statistical
terms, however, the annual data series of GDP can be considered a combination of three
processes; a long-run trend, business cycles, and short-run shocks to the economy, which can be
separated from each other by using statistical techniques. The objective of this paper is to
decompose real Gross Domestic Product of Pakistan into its above-mentioned components, and
also to project them for determining future path of real GDP. Such type of dialysis gives more
insights to understand the changing pattern of economic growth. Although some very broad
comments have been made in the paper on the behaviour of GDP components during the past
decades, a more detailed investigation is left for future research1.
In the literature, a number of methods have been proposed for separating the trend from the
cyclical component of an economic time series. The most popular of these is the Hodrick-
Prescott (1997); while others include Rotemberg (1999), Baxter-King filter (1995), etc. This
study has used the Hodrick-Prescott filter2. A brief summary of the methodology used is
provided in the following section. Results are presented and explained in section 3, while the
last section concludes the paper.

2. Methodology
The methodology consists of two steps; first to dissect the real GDP to get its components, and
second to project the components into future. It is assumed that annual series of real GDP is an

1
The scope of this paper is limited to decompose the real GDP statistically and to project it into future. However, the
results of the study raises a number of unanswered questions and pave way for future research in this area which may
include use of alternate methods of decomposition, co-integration between trend and cycles exist in GDP,
consumption, investment, M2, etc, and of course a detailed explanation of shocks, cycles and trends with respect to
structural, political, social, and global events.
2
See Pedregal & Young (2000), Pederson (1998), and Reeves et all (2000) for some useful comments on HP Filter.

3
aggregate of three components viz. Trend, Cyclical movements, and Irregular movements3. In
symbolic form;

Yt = Tt + Ct + It .1

where Tt = Long-run trend

Ct = Cyclical movements
It = Irregular movements (shocks)
All are in natural log form.

The HP filter is used in two stages to separate these components; first to extract the long-run
trend (Tt) from the original series and then to filter out cycles (Ct) from the rest. The HP filter
proceeds as follows;
It assumes that a series (say Xt) has two components; a smooth one (St) and deviations (Dt) from
St, i.e.
Xt = St + Dt .2
such that over a long period of time the sum of deviations (Dt) is near zero.
In order to filter out St from Xt, it minimizes the following:
Min: Σ Dt2 + λ Σ [∆2St] 2 .3
The parameter λ is a positive number, which penalizes variability in the smooth component (St).
∆ is the difference operator, power of which shows the order of differentiation. The higher the
value of λ, the smoother is the solution series. Hodrick and Prescott suggested λ = 100 for
annual data.
We have applied the HP filter on the original series of real GDP to extract trend (Tt) component
from it. By subtracting the trend from the original series (Yt), we get a new series (Zt) that
contains cyclical and irregular components.
Zt = Yt – Tt = Ct + It .4

3
This assumption is similar to that adopted by U.S. Bureau of Census in its seasonal adjustment program (the latest
version of which is X-12-ARIMA). In a seasonal adjustment program, a series is usually assumed to have four
components: long-run trend, business cycle, seasonal variations, and irregular movements. However, in case of
annual data there will be no seasonal component, leaving only three types of variations.

4
We again apply the HP filter on Zt. In this second stage the HP filter wheedles out oscillations
around the smooth component that is nothing but Cycles Ct. The difference between Zt and Ct
represents shocks or Irregular component (It)
The next step is to project trend and cycles into future over a ten-years’ period. For this purpose,
ARIMA models have been used. Following the usual procedure of selecting a time series
model4, we have identified the following models for components;

For trend component (Tt)5:

(1-α1L-α2L2 -α3L4) (1-L)2 Tt = α0 + (1+β1L+β2L4)ε .5

For cyclical component (Ct)

(1-φ1L-φ2L2 -φ3L3-φ4L4) Ct = φ0 + (1+π1L+π2L2+π3L3) ε .6

Data of real GDP at prices of 1980-81 for years 1949-50 to 2000-01 has been used6.

3. Results
The three components of Pakistan’s real GDP have been shown in Figure-1. As expected, it has
a positive long-run trend throughout the period. However, it may be noted that though trend
GDP is increasing over time, its rate of increase as depicted in Figure-2 has different behaviour
in different periods. During the first half of 1960s, the growth of trend component was rising
while the second half witnessed a slow down. This pattern of growth has some bearings on
structural developments in early 60s including the green revolution, the industrial revolution,
the development of financial institutions, etc., and the impact of 1965 war on the later half of
the 60s.

4
The procedure includes establishing order of integration by unit root test, identification of model with the help of
autocorrelation and partial autocorrelation functions, and diagnostics etc.
5
L is lag operator and ε is error term, normally distributed around constant mean.
6
Data source for 1949-50 to 1994-95 is ’50 Years of Pakistan, Volume 1 – Summary, by Federal Bureau of Statistics
(June 1998); for 1995-96 onward Economic Survey – 2000-01, by Economic Affairs Division, Ministry of Finance
(June 2001).

5
Figure 1: Components of Real GDP
14

13.5
Long-run Trend
13

12.5

12

11.5

11

10.5

10

0.02

0.015
Business Cycle
0.01

0.005

-0.005

-0.01

-0.015

-0.02

0.05

Short-run Shocks
0.03

0.01

-0.01

-0.03

-0.05

Note: Unit of Y-axis is natural log of million rupees.

6
 Figure 2: Trend Growth rate (%) & Public Investment (% of GDP)
% grow th % of GDP
6.5 14
12.6 % 12.4 %
6 12

5.5 9.4 % 10
9.2 %

5 8

4.5 6

4 4
Grow th of Trend (%)
Average public investment (% of GDP)
3.5 2

3 0
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
Another very intriguing result is the increasing trend growth during 1970s – a poorly rated
decade by economists. There may be more explanations to it; one is the role of public fixed
investment. Public fixed investment was 12.4 percent of GDP during 1973-74 to 1983-84
compared with 9.7 percent and 9.4 percent during the two adjacent periods (Figure-2). It
suggests that some positive structural changes occurred during this period. However, as
mentioned below, the 1970s also witnessed a recession of second business cycle, which may
have some fainting effects on the overall economic performance.
Since 1983-84, the trend growth is declining, despite high overall GDP growth of 1980s. It
suggests that high growth of 1980s may have been due largely to some transitory factors,
including large inflow of remittances and foreign aid, instead of structural changes in the
economy that could support long run growth.
As regards the cyclical movements, it is shown that real GDP in Pakistan has completed at least
two business cycles; one ending with a peak in 1968-69, and a second ending in 1990-91. Since
1991-92, the economy is in recessionary phase of a third business cycle. The following table
gives a time frame of phases of business cycles in Pakistan.

7
Table 1: Time Frame of Business Cycles in Pakistan

Business Cycle Recession Trough Recovery Peak

First B. Cycle 1949-50 to 1959-60 1959-60 1960-61 to 1968-69 1968-69
(1949-69) (11 years) (9 years)
Second B. Cycle 1969-70 to 1978-79 1978-79 1979-80 to 1990-91 1990-91
(1969-1991) (10 years) (12 years)
Third B. Cycle 1991-92 to 2004-05 * 2004-05 * 2005-06 to ?
(1991 - ?) (14 years)
* Estimated

The above chart reveals the following features of business cycles in Pakistan.

1. Pakistan’s economy went into recession soon after the independence. The whole decade
of 1950 witnessed economic sluggishness that may be attributed to communal upsets,
lack of infrastructure, weak (or virtually absent) industrial base, lack of private sector
confidence on the infant economy, etc.
2. The economy started recovering by 1960. Interestingly, the recovery period is shorter
than the recession. It may be postulated that appropriate economic planning and its
effective implementation helped the economy recover quickly.
3. During 1970s the economy fell into recession almost as quickly as it had recovered
during the last decade. Separation of East Pakistan and the nationalization of industrial,
financial and other institutions could have adversely affected the business confidence
during this period.
4. It took 12 years for the economy to recover from second recession compared with a 9-
year recovery period in the first cycle. Particularly, the economy slowed down in late
1980s, and did not achieve even the peak level of last cycle that it fell into next
recession.
5. Third recession started in early 1990s and is projected to continue till 2004-05.
According to our projection, it will be the deepest recession of all. (Figure-3).

8
Figure 3: Projections of Trend and Business Cycle
14 7
Long-run Trend
13.5 6

13
5
12.5 Projected
4

% grow t
12
% Grow th in trend 3
11.5
2
11

10.5 1
Trend

10 0
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
0.02

0.015
Business Cycles
0.01
Projected
0.005

-0.005

-0.01

-0.015

-0.02

12

Overall GDP Grow th (%)

10

Projected
6

-2

Note: : Unit of primary Y-axis is natural log of million rupees

9
Figure-3 also shows the projection of long-run trend. It is Table2: Projected real GDP

estimated that the trend growth will start rising from year 2001- Growth

02. Although the cyclical component of the GDP is projected to Year % Growth
2001-02 4.251
decline further in coming years, the dominance of trend will
2002-03 3.528
drive the overall GDP growth up (Table-2) 7. The projection is
2003-04 3.692
based on the assumption that the economy will not suffer from
2004-05 3.897
positive or negative shocks during that period.
2005-06 4.111
2006-07 4.308
Figure-4 shows the irregular component of the GDP that
2007-08 4.464
represents short-run shocks to the economy. Pakistan’s economy 2008-09 4.564
has witnessed number of positive and negative shocks during the 2009-10 4.601

Figure 4: Irregular Component of GDP and Cotton Production

0.05 14

0.04
12
0.03

10
0.02

0.01
8

6
-0.01

-0.02 4

-0.03
Short-run Shocks 2
-0.04
Cotton production
-0.05 0

Primary Y –axis = natural log of million rupees; Secondary Y-axis = million bales

past decades. Although explaining each and every shock is outside the scope of this study, one
important source of shocks is identified as cotton production in Pakistan. When there was a
bumper cotton crop, the growth rose sharply (as in 1992 and 1996); and when it failed, growth

7
The projection shows a sharp jump in GDP growth during 2001-02, which comes down then to the onset
of a smooth path. The reason may be very nature of the time series models, which keep in memory the
past data generating process. Sharp rise in growth during 2001-02 is possibly the lagged impact of some
previous process, which is smoothed out subsequently.

10
was depressed. A comparison of cotton production and irregular component of the GDP
(Figure-4) clearly shows co-movements8 of both the series indicating the power of cotton of
driving the real GDP above/below its trend and cyclical path.

4. Conclusion
The paper has decomposed, statistically, the real GDP of Pakistan into three components viz.,
long-run trend, business cycles and short-run shocks. It also projected GDP growth on the basis
of projected trend and business cycle. It is found that trend growth of real GDP, though positive,
is declining since early 1980s. The reversal of trend growth is however, expected from year
2001-02. The results also show that the economy of Pakistan has undergone two complete
business cycles since its independence, and is now facing the recessionary phase of a third
business cycle, which began in the early 1990s. It is projected that the current recession will
continue until 2004-05, and then recovery will take place. However, since the dominant
component of real GDP, the trend is projected to gain strength from 2001-02, the overall GDP
growth is projected to rise gradually in coming years. The paper identifies a number of positive
and negative shocks to the economy during past decades, and also finds a significant correlation
between cotton production and shocks.

8
Correlation coefficient of cotton production and irregular component is +0.21.

11
References
Baxter, M & King R.G. (1995); Measuring business cycles approximate Bank-Pass filters for
economic time series, NBER, WP 5022, February 1995

Box G.E.P. & Gwilyn M. Jenkins (1970); Time series analysis, forecasting and control, San
Francisco, Holden Day, 1970

Hodrick, R. J. & Edward Prescott (1997); Post-war business cycle: an empirical investigation,
Journal of Money, Credit & Banking 29 (1), February 1997

Lucas, Robert E. (1981); Studies in business cycle theory, Cambridge, Mass; The MIT Press,
1981

Rotemberg, J.J. (1999); A heuristic method for extracting smooth trends from economic time
series, NBER, WP 7439, December, 1999.

Reeves J.J., Blyth A. C., Triggs M. C, & Small P.J. (2000); The Hodrick-Prescott Filter, a
Generalization and a New Procedure for Extracting an Empirical Cycle from a Series; Studies in
Nonlinear Dynamics and Econometrics, April 2000, 4(1): 1-16

Pederson M. Torben (1998); The Hodrick-Prescott Filter, the Slutzky-effect and the
distortionary effect of filters; DP No. 9/1998; Institute of Economics, University of Cpenhagen,
November, 1998

12