SECTORAL ANALYSIS OF THE RELATIONSHIP BETWEEN STOCK RETURNS, OUTPUT GROWTH AND INFLATION

1
SECTORAL ANALYSIS OF THE RELATIONSHIP BETWEEN
STOCK RETURNS, OUTPUT GROWTH AND INFLATION
Nildag Basak CEYLAN (Atilim University,
Department of Management) Sidika BASÇI (SESRTCIC)
Ankara 30 January 2007
2

• Boudoukh, Richordson and Whitelaw (1994)
• Catogarized the industries as cyclical and
  noncyclical where cyclical industries are the
  ones which have outputs highly correlated with
  the aggregate output. The reverse is true for the
  nocyclical industries.
• They show that there is a positive relation
  between stock returns and expected inflation for
  the non-cyclical industries and a negative one
  for the cyclical industries for US market.
• In this study, we examine the same issue for
  Turkish Stock Market.

3

• METHODOLOGY (I)
• Determination of Cyclical and Non-cyclical
  Sectors
• By calculating the correlation coefficients for
  industries output growth and aggregate output
  growth.
• By regressing every industrys output growth on
  aggregate output growth and find the coefficient
  beta.

• where is the aggregate output growth,
  is the output growth
• of the ith sector at time t.
• if isgt1 then the industry is cyclical and if
  lt1 then the industry
• is non-cyclical

4

• METHODOLOGY (II)
• Relationship Between Industry Stock Returns and
  Inflation
• where is the stock return of sector i,
  is the expected inflation and
• is the unexpected inflation at time t.
• We calculate the expected inflation by regressing
  inflation data on its own lag values and current
  auction interest rate.
• The lags that should be included in the model are
  determined by Hendrys General to Specific
  Methodology.
• The unexpected inflation is the residuals
  obtained from this regression.

5

• DATA
• Our sample period is January 1997 to November
  2006.
• We got the output data of industries from Turkish
  Statistical Institude (TURKSTAT) and stock market
  price indices of industries from datastream.
• While doing this we chose the industries where
  both of the sources include. As a result we had
  to study only for eight industries.
• We took the first logarithmic differences of
  indices of industries.
• For aggregate output (GDP) and auction interest
  rate, the data are gathered from the EVDS system
  of CBT and the data of industrial production
  index is obtained from IFS.
• The data are seosonally adjusted, and to
  calculate the growths, we took the first
  differences of each series.
• Monthly and quarterly data of inflation are also
  taken from the EVDS system of CBT.

6
Table 1 This table reports the correlation
between output growth of eight industries and
growth of aggregate output correlation between
these industries output growth and inflation.
Aggregate output growth is measured by the change
of Industrial Production Index. Beta is the
coefficient from the regression of industry
output growth on aggregate output growth. The
values in pharanthesis are standard deviations.
The sample covers seasonally adjusted monthly
data over the January 1997 to November 2006
period.
7
Table 2 This table reports the correlation
between output growth of eight industries and
growth of aggregate output correlation between
these industries output growth and inflation.
Aggregate output growth is measured by the change
of Industrial Production Index. Beta is the
coefficient from the regression of industry
output growth on aggregate output growth. The
values in pharanthesis are standard deviations.
The sample covers seasonally adjusted quarterly
data over the 1997 to 2006 period.
8
Table 3 This table reports the correlation
between output growth of eight industries and
growth of aggregate output correlation between
these industries output growth and inflation.
Aggregate output growth is measured by the change
of Gross Domestic Product. Beta is the
coefficient from the regression of industry
output growth on aggregate output growth. The
values in pharanthesis are standard deviations.
The sample covers seasonally adjusted data over
the 1997 to 2006 period.
9
Non-cyclical industries, when monthly data is
used are Primary metals Food and
beverage Non-cyclical industries, when quarterly
data is used are Metal products and
machinery Textile and leather Lumber,
paper, printing- publishing and furniture
Food and beverage Primary metals
10
The model for expected inflation obtained from
Hendrys methodology both for the monthly and
the quarterly data is as follows
11
Table 4 This table reports the coefficients from
the regression of industry stock returns on
expected and unexpected inflation. Values in
pharantesis are standard deviations. The sample
covers monthly data over the January 1997 to
November 2006 period. Expected inflation is based
on a time-series model using past inflation.
12
Table 5 This table reports the coefficients from
the regression of industry stock returns on
expected and unexpected inflation. Values in
pharantesis are standard deviations. The sample
covers quarterly data over the 1997 to 2006
period. Expected inflation is based on a
time-series model using past inflation.

13

• CONCLUSION
• Non-cyclical industries, when monthly data is
  used are
• Primary metals
• Food and beverage
• Non-cyclical industries, when quarterly data is
  used are
• Metal products and machinery
• Textile and leather
• Lumber, paper, printing- publishing and
  furniture
• Food and beverage
• Primary metals
• We do not see perfectly that for non-cyclical
  industries there is a positive relation between
  stock returns and expected inflation and there is
  a negative relation for cyclical industries like
  the case of US.

14

• As a future work
• Instead of IP growth expected IP growth can be
  used.
• The following model can be used in the
  determination of expected IP growth

• While calculating the estimated value of IP,
  capacity utilization
• rate can be used.