Proceedings of 21st International Business Research Conference
10 - 11 June, 2013, Ryerson University, Toronto, Canada, ISBN: 978-1-922069-25-2
The Dynamic Relationship between Economics and
Financial Instability
H. Feyza Erdem* and Rahmi Yamak**
After 1980s, the long-term macroeconomic instability in Turkish
Economy has become a serious problem within financial
liberalization. Recent institutional changes in Turkish financial
market drove us to re-examine the dynamic relationship between the
financial instability and macroeconomic instability. The aim of this
study is to investigate the causal relationship between economic
instability and financial instability for Turkish Economy. The data
used in the study are quarterly and cover the period of 1998-2012. In
this study, GDP and M2Y/GDP were used as the macro-economic
indicator and financial indicator, respectively. Kalman Filter
Technique among various algorithmic approaches was used to get
economic and financial instability indicators. After getting both
indicators, Box-Jenkins models of both variables were statistically
estimated by using Kalman Filter. The relationship between
economic instability and financial instability was examined by using
Granger Causality test. Consequently, in the short-term, economic
instability causes financial instability and financial instability causes
economic instability.
JEL Codes: E32, C53, C32, E20
1. Introduction
Central Banks in many countries, especially in high- and medium-income countries
have focused the stabilities of financial and economic sector. Financial and
economic stabilities have emerged as an important public policy objective. Recently
almost all economies have got economic and financial instabilities. Central Banks in
all countries have become more concerned about what leads to instability and what
can be done to prevent it.
There are several negative effects of economic and financial instabilities. The most
important one is that economic instability creates a negative impact on economic
decisions of economic units. Uncertainty associated with an unstable economic
environment may reduce production, increase unemployment and demolish income.
One of the important effects of economic instability; it negatively affects finance
markets. Instability in financial markets may lead to financial crisis and
macroeconomic imbalance. Under these circumstances, financial instability has been
appeared as a remarkable problem especially for developing countries.
According to Mishkin (1997), although financial instability has struck industrialized
countries just as frequently, a particularly severe problem for emerging-market
countries suffer disproportionately when it occurs.
*Research Assist. H. Feyza ERDEM, The Department of Econometrics, Karadeniz Technical
University, Trabzon / Turkey, Email: havvanurerdem@ktu.edu.tr
**Prof. Dr. Rahmi YAMAK, The Department of Econometrics, Karadeniz Technical University,
Trabzon/ Turkey, Email: yamak@ktu.edu.tr
1
Proceedings of 21st International Business Research Conference
10 - 11 June, 2013, Ryerson University, Toronto, Canada, ISBN: 978-1-922069-25-2
In the last three decades, Turkey has had many political, economic and financial
instabilities. However, recently Turkish economy has experienced more stable
economic and financial conditions. The events of the last years have posed
challenges for policymakers and have motivated to re-examine the relationship
between financial and economic instability in Turkish economy. This paper
addresses two questions: Does economic instability cause financial instability, or
does financial instability lead to economic instability?
Within this framework, the purpose of this study is to investigate the causal
relationship between economic instability and financial instability for Turkish
Economy. The data used in the study are quarterly and cover the period of 19982012. Two important features arise from current study. First, we examine the
possible dynamic relation between economic instability and financial instability.
Second, our study differs from other studies by using Kalman Filter Technique to get
economic and financial instability indicator. We use GDP and M2Y/GDP as the
macro-economic indicator and financial indicator, respectively.
The study is organized as follows: Firstly, the theoretical and empirical literature is
discussed in terms of the relationship between economic instability and financial
instability. Secondly, data is described and the methodology is given. Thirdly, the
causal relationship between economic instability and financial instability is analyzed
by presenting empirical findings. Finally, we summarize econometric findings.
2. Literature Review
Economic and financial instability has been subject to many studies. In the related
literature, there are two important questions: why and how economic instability may
play a crucial role in the financial sector and why and how financial instability may
affect the whole economy? There are several approaches to investigate the analysis
of the dynamic behaviors of the economic and financial system. It is useful to begin
by summarizing the basic intuition underlying the theory of economic and financial
instability, and some of the more important results from the literature.
One of the most important approaches has been the „Financial Instability Hypothesis‟
which was proposed by Hyman Minsky (1919 – 1996). „Financial Instability
Hypothesis‟ mean that the financially-dominated capitalist economy has been
inherently unstable (Minsky 1975, 1982, 1986). According to Minsky, two theorems
of the financial stability hypothesis: (1) “the economy has financing regimes under
which it is stable, and financing regimes in which it is unstable”, and (2) “over periods
of prolonged prosperity, the economy transits from financial relations that make for a
stable system to financial relations that make for an unstable system.” Minsky, based
his theory on the inbuilt instability of financial markets on Keynes‟ theory of money, in
which the key role was played by the explanation about the speculative demand for
money.
Following Keynes‟ teaching, he was an advocate of state intervention on financial
markets, sharply opposing the deregulatory measures that were applied to these
markets during the 1980s in the US. (Jovancai, 2010).
2
Proceedings of 21st International Business Research Conference
10 - 11 June, 2013, Ryerson University, Toronto, Canada, ISBN: 978-1-922069-25-2
Minsky‟s theory has been investigated on instability of financial markets among
economists. For example, Kindleberger (1989) was studied about financial instability,
based Minsky hypothesis. He explained “market euphoria”. According to
Kindleberger, financial instability had five steps: displacement, boom,
overinvestment, revulsion and tranquility. Friedman (1982), the interventions in the
supply of money and credit channels can cause economic instability. In other words,
the central poor management of banks, the money supply increases or the extreme
narrowing policies, can lead to the disruption of financial markets and crises.
Bockelmann and Borio (1990) investigated financial instability and real economy.
The paper deal with three different areas of the financial sphere: national moneys,
exchange rates and financial claims, markets and institutions. Pindyck and Solimano
(1990) examined economic instability and aggregate investment. They explored the
empirical relevance of irreversibility and uncertainty for aggregate investment
behavior.
Mishkin (1999) examined what causes and propagates financial instability. Mishkin
used to understand the structure of the financial system and he suggested that there
were four categories of factors that lead to financial instability: increases in interest
rates, increases in uncertainty, asset market effects on balance sheets, and
problems in the banking sector. Martin and Rogers (2000) examined long-term
growth and short-term economic instability. Brousseau and Detken (2001) presented
an example of a possible conflict between short-term price stability and financial
stability. Allen (2005) studied modeling financial instability. Lima and Meirelles
(2006), the financial fragility of the economy explained on the modeled Minsky.
Jeanneney and Kpodar (2011) investigated financial instability which accompanied
financial development was detrimental to the poor and dampened the positive effect
of financial development on the reduction of poverty.
3. Data and Methodology
The data used in this study were quarterly and cover the period from 1998:012012:01 for Turkey. Real Gross Domestic Product (GDP) and ratio of broad money
supply to GDP (M2Y/GDP) are used as the macro-economic indicator and financial
indicator, respectively. The both variables are obtained from Electronic Data Delivery
System, the Central Bank of the Republic of Turkey (TCMB_EVDS). The both
variables are adjusted seasonality1 and are used to be logarithm.
The econometric process used is followed in this way: first of all, as an empirical
enquiry, we deal unit root tests procedures to determine whether GDP and
M2Y/GDP variables are indeed stationary. We use two different unit root tests to
determine whether the GDP and M2Y/GDP series are stationary: developed by
Dickey and Fuller (1979) (Augmented Dickey-Fuller (ADF))2 and by Phillips and
Perron (1988) (PP)3. After then, getting both indicators, Box-Jenkins models of both
variables are statistically estimated by Kalman Filter Technique. Since Kalman Filter
Technique enables time varying error term variations to be obtained, the time varying
error term variation is used as a criterion for instability series. Finally, the probable
1
2
3
We used Moving-Average Methods to adjust seasonallity. For more information: Winters (1960).
For more information: Dickey, D. and Fuller, W. (1979, 427-431).
For more information: Phillips, P. and Perron, P. (1998, 335-346).
3
Proceedings of 21st International Business Research Conference
10 - 11 June, 2013, Ryerson University, Toronto, Canada, ISBN: 978-1-922069-25-2
relationships between the economic and financial instability series are examined by
using time series techniques.
In the Kalman Filter estimation technique, the first necessary step is to construct the
state space form, which consists of measurement and transition equations (Kalman,
1960). Measurement equation is not different of standard OLS regression equation‟s
coefficient which is added time factor. The following equation (1) is measurement
equation.
t
+
t
t t
+
t
E( t )=0 and V( t )= Vt
(1)
The transition equation is the system of equation how changing parameters of
measurement equation change depending over time. In this studying, it was
assumed that variable parameters of measurement equation has AR(1) structure.
According to (1) number equations, there are two transition equations.
t
t
t1
t2
t-1
t-1
+
+
(2)
(3)
1t
2t
To explain Kalman Filter process, it must be expressed (1), (2), (3) equations by
matrix form. (4) and (5) equations are matrix form of (1), (2) and (3) equations.
yt xt
t
+
t
t-1
+
t
(4)
(5)
t
(4) equation is expression as matrix of (1) measurement equation. While y
represents Y, x does X (including the constant term). The software in the form of the
transition equation is (5) equation. Z represent the vector of size 2x1 that has
elements and , represent main diagonal. t1, t2 represent the matrix of size 2x2
which is zero off-main diagonal and t , describe the vector of size 2x1 that has
elements 1 , 2 .
In the first step, by using the initial or unconditional estimates of Z and their variancecovariance matrix P, the conditional estimates of Z and their conditional variancecovariance matrix are obtained from the following equations (6) and (7).
Zt|t-1 =
Pt|t-1 =
(7)
(6)
t-1
t-1
+
In the second step, the conditional y, the one step ahead prediction error H, and its
conditional variance F, are estimated by using outputs of the first step and the
following equations (8) - (10).
4
Proceedings of 21st International Business Research Conference
10 - 11 June, 2013, Ryerson University, Toronto, Canada, ISBN: 978-1-922069-25-2
yt|t-1 = xt Zt|t-1
Ht = yt - yt|t-1
Ft = xt Pt|t-1xt ' + V
(8)
(9)
(10)
In the final step, the unconditional Z and its variance-covariance matrix P, are
obtained by utilizing the outputs of the previous steps and the following updating
equations (11) and (12).
'
t =Pt|t-1–(Pt|t-1xt Ft
-1
xt Pt|t-1)
(11)
'
-1
t = Zt|t-1 + Pt|t-1xt Ft Ht
(12)
Once the filter completes all three steps and provides unconditional P and Z, then
the unconditional estimates enter into step 1, as being inputs and the filter again
starts to work to complete all three steps for t+1 and continues until last time period,
t-1. Therefore, the Kalman Filter is known to be a recursive estimation technique
through time.
4. Empirical Findings
Before we estimate the system that establishes the relationship between economic
and financial instabilities, we summarize results of unit root tests (ADF and PP) for
the level and first difference of both variables in Tables 1 and 2.
Table 1: Unit Root Test Results for GDP
Exogenous
ADF
PP
GDP
ΔGDP
GDP
ΔGDP
Intercept
-0.1645 -7.5307 ***
0.4552
-8.8726 ***
Trend and Intercept
-2.7482
-7.6094***
-2.6812
-10.5620 ***
None
2.0102
-6.5923 ***
3.5104
-6.5600 ***
Note:*** %1 level test critical values. Δ: first difference of the variable.
Exogenous
Table 2: Unit Root Test Results for M2Y/ GDP
ADF
PP
Intercept
Trend and Intercept
None
M2Y/ GDP
Δ (M2Y/ GDP)
M2Y/ GDP
-2.6154*
-2.5466
0.4450
-5.4621***
-5.5887***
-5.2200***
-2.5047
-2.4923
0.1086
Δ (M2Y/
GDP)
-5.4705***
-5.6002***
-5.2115***
Note:*** %1 level, **%5 level, *%10 level test critical values. Δ: first difference of the variable.
The null hypothesis for ADF and PP tests indicate the existence of a unit root.
According to the unit root test results, GDP and M2Y/GDP is stationary in the first
difference. So, ADF and PP tests reject the null hypothesis of a unit root in the series
at the 0.01 level, indicating that the both series are stationary in the first differences.
In order to obtain instabilities of economic and financial series, first differences of the
both series are used.
5
Proceedings of 21st International Business Research Conference
10 - 11 June, 2013, Ryerson University, Toronto, Canada, ISBN: 978-1-922069-25-2
Before we get series of economic and financial instability under the Kalman Filter
techniques, firstly, we must determine the best ARIMA4 model for GDP and
M2Y/GDP. We decided the best model as ARIMA(2,1,2) and ARIMA(2,1,1) for GDP
and M2Y/GDP series, respectively. In Tables 3 and 4, the results of the best ARIMA
models are given for both series.
Table 3: ARIMA(2,1,2) Model for GDP
Coefficient
Std. Error
t-Statistic
C
0.0106
AR(1)
-0.0758
AR(2)
-0.9164
MA(1)
0.2898
MA(2)
0.9604
2
R
0.40
Inverted AR Roots -.04-.96i
Inverted MA Roots-.14+.97i
0.0047
0.0396
0.0438
0.0281
0.0285
2.2356
-1.9145
-20.8842
10.3126
33.61215
Prob.
0.0300
0.0614
0.0000
0.0000
0.0000
-.04+.96i
-.14-.97i
Note: In order to find out whether the residuals are serially correlated, Breush-Godfrey Serial Correlation Lagrange Multiplier
(LM) test were used for first and fourth order serial correlation. LM1: 0.38 and LM4: 0.93. There were no correlated.
Table 4: ARIMA(2,1,1) Model for M2Y/ GDP
Coefficient
Std. Error
t-Statistic
C
AR(1)
AR(2)
MA(2)
0.0081
1.1260
-0.3086
-0.9591
R2
Inverted AR Roots
Inverted MA Roots
0.0025
0.1388
0.1300
0.0513
0.21
.65
.96
3.2085
8.1090
-2.3727
-18.6786
Prob.
0.0023
0.0000
0.0215
0.0000
.47
Note: For first and fourth order serial correlation, Breush-Godfrey test‟s results: LM1: 0.94 and LM4: 0.74. There were no
correlated.
2
As seen in Tables 3 and 4, R statistics of ARIMA models are 0.40 and 0.21,
respectively. AR Roots <|1| and MA Roots < |1| for both models. These mean that
estimated AR processes are stationary and estimated MA processes are invertible.
In addition, the estimated coefficients are statistically significant at %10 level for both
models.
After Box-Jenkins models are estimated for both variables, the Kalman Filter
Techniques is run and economic and financial instabilities are derived. In Figures 1
and 2, economic and financial instabilities graphics are shown.
4
We computed the correlograms of GDP and M2Y/GDP series to define which ARIMA model was
best. We run various ARIMA models with different orders: ARIMA(1,1,1), ARIMA(1,1,2),
ARIMA(1,1,3), ARIMA(2,1,1), ARIMA(2,1,2), ARIMA(2,1,3), ARIMA(3,1,1), ARIMA(3,1,2),
ARIMA(3,1,3). For more information about ARMA models: Box, G.E.P. and G.M.Jenkins (1976, 575).
6
Proceedings of 21st International Business Research Conference
10 - 11 June, 2013, Ryerson University, Toronto, Canada, ISBN: 978-1-922069-25-2
Figure 1. Economic Instability
Figure 2. Financial Instability
Figure 1 reveals that economic instability takes on its highest value at the fourth
period of 1998 and on its lowest value at the first period of 2010. Figure 2 shows that
the lowest value of financial instability exists at the second period of 2009 and the
highest value exists at the fourth period of 2001. Especially, the high values of
economic and financial instabilities are accompanied at the crisis periods: 1998,
2000, 2001, 2008, 2010, 2011.
As a final step, we investigate the relationship between economic and financial
instability variables. For this purpose, Granger causality5 test is conducted, which
essentially test whether there is a consistent lead and lag relationship between the
both variables. The Granger causality between the two variables, economic
instability and financial instability, asks that how much of the current economic
instability can be explained by a regression on its past values, and then tries to test
whether inclusion of the lagged values of financial instability into the regression to
explain economic instability have statistical significance as a whole.
Table 5 presents the Granger causal relation between economic and financial
instabilities.
Table 5: Granger Causality Results
Equations
EcoInst.(p)
FinInst (p)
LM(1)
R2
FTEST(Wald)
(1)EcoInst.→FinInst.
4
8
0.3916
[0.5325]
4. 8162
[0.0.005]
EcoInst causes FinInst.
0.57
(2)FinInst.→EcoInst.
8
4
0.0039
[0.9505]
FinInst causes EcoInst.
0.54
3.2245
[0.0236]
Result
Note: p is optimal lag for AIC. For fourth order serial correlation, Breush-Godfrey test were tested for the equations. According
to these results, there is no fourth-order correlation. [ ] expresses probability values. First Equation: LM4 : 0.2454 [0.9102],
Second Equation: LM4 : 0.2542 [0.9042]. . EcoInst. is economic instability variable and FinInst. is financial instability variable
The null hypothesis in Equation (1) is that the lags of economic instability is not
significant as a whole, that is to say, economic instability does not Granger cause
5
For more information of Granger Causality: Granger, C. W. J. (1969, 424-438). Unit-root
tests(Augmented Dickey-Fuller and Phillips-Perron) of the instability series are tested. According to
unit root test results, both instability series are stationary in the level.
7
Proceedings of 21st International Business Research Conference
10 - 11 June, 2013, Ryerson University, Toronto, Canada, ISBN: 978-1-922069-25-2
financial instability. Likewise, the null hypothesis in Equation (2) is that the lags of
financial instability have no statistical significance in explaining economic instability
which also means that financial instability does not Granger-cause economic
instability. By employing F-type Wald tests, the results of pairwise Granger causality
analysis which are applied on the joint significance of the sum of lags of each
explanatory variable are reported below. For this purpose, various lag lengths are
considered to see whether the estimation results are sensitive to the a priori lag
selection. The results of F-type Wald tests indicate significance at the 0.05 and 0.10
levels for both equations. So, we can easily notice that the null hypotheses are
rejected at the 0.05 and 0.10 significance level for both equations. In other words, we
find that economic instability causes financial instability and financial instability cause
economic instability. There are two-sided relation between economic instability and
financial instability in Turkish economy in short-term.
5. Conclusions
Stability of a country has been widely evaluated in a multi-dimensional perspective
and in a long-term process. In this process, national and international results of a
country‟s stability level are usually interpreted as an indicator of the performance in
national and international arena. For all developing and developed economies,
macroeconomic stability is not independent from financial stability. Nowadays,
almost all economies have stability problems in the financial markets where there are
predictable and unpredictable changes within globalization. There are several
negative effects of instability in financial markets, such as reduction of production,
increase of unemployment, demolish of income and resources distribution, financial
crisis, macroeconomic imbalance. After 1980s, the long-term macroeconomic
instability in Turkish Economy has become a serious problem within financial
liberalization.
In this study, the causal relationship between economic instability and financial
instability were examined for Turkish Economy. The data used in the study are
quarterly and cover the period of 1998-2012. Due to the fact that instability series
was not directly observable, economic instability and financial instability variables
were obtained by using Kalman Filter technique. Economic and financial indicators
were described, as GDP and M2Y/GDP, respectively. After getting both indicators,
Box-Jenkins models of both variables were statistically estimated by Kalman Filter
technique. Then, the economic and financial instability were obtained by Kalman
Filter technique. After getting both instability series, a probable the relations between
economic and financial instabilities variables were analyzed by using Granger
Causality. As a result, in the short-term, economic instability causes financial
instability and financial instability causes economic instability. So, the causal
relationship between economic instability and financial instability were get as twosided.
References
Allen, F. 2005. “Modeling financial instability, National Institute Economic
192, pp. 57-67.
eview”,
8
Proceedings of 21st International Business Research Conference
10 - 11 June, 2013, Ryerson University, Toronto, Canada, ISBN: 978-1-922069-25-2
Bockelmana, H. and Et Borio, C. 1990. “Financial instability and the real economy,
De Economist”, Vol.138, N4, pp. 428-450.
Box, G.E.P. and Jenkins, G.M. 1976. “Time series analysis: forecasting and control.
holden-day”, San Francisco, 575.
Brousseau, V. and Brousseau, V., Detken, C. 2001. “Monetary policy and fears of
financial instability, European Central Bank”, Working Paper.
Dickey, D. and Fuller, W. 1979. “Distribution of the estimators for autoregressive time
series with a unit root, Journal of the American Statistical Association”, 74,
ss.427-431.
Friedman, M. Schwartz, A. J. 1982. “The role of money, National Bureau of
Economic esearch”, 0-226-26409-2, 621 – 632.
Granger, C. W. J. 1969. “Investigating causal relations by econometric models and
cross spectral methods, Econometrica”, 37, 424–438.
Jeanneney S. G. and Kpodar, K. 2011. “Financial development and poverty
reduction: can there be a benefit without a cost?, The Journal of Development
Studies”, 47.
Jovancai A., M.A. 2010. “Instability of financial markets as a cause of the global
financial crisis the minsky moment, Scientific Review Paper”, Vol. 7 (1) 2010:
pp. 17-28
Kalman, R. E. 1960. “A new approach to linear filtering and prediction problems,
Journal of Basic Engineering”, 82:34-45.
Kindleberger, C., 1989. “Manias, panics, and crashes: a history of financial crises,
New York: Basic Books”, Inc.
Lima, G.T. and Meirelles, A.J.A. 2006. “Debt, financial fragility, and economic
growth: a post keynesian macromodel, Journal of Post Keynesian Economics”,
29: 93-115.
Martina, P. and Rogers, C.A. 2000. “Long-term growth and short-term economic
instability, European Economic Review”, Volume 44, Issue 2, 359–381.
Mishkin, F.S. 1997. “The causes and propagation of financial instability: lessons for
policymakers, In Federal Reserve Bank of Kansas City, Maintaining Financial
Stability in a Global Economy”.
Minsky, H. P. 1975. “John Maynard Keynes and New York: Columbia University
Press”.
Minsky, H. P. 1982. “Can "it" happen again?, Essays on Instability and Finance.
Armonk and N.Y.: M.E. Sharpe”.
Mishkin, F.S. 1999. “Global financial instability:framework, events, issues, Journal of
Economic Perspectives”.
Minsky, H.P. 1986. “Stabilizing an unstable economy, New Haven, Conn.:Yale
University Press”.
Phillips, P. ve Peron, P. 1988. “Testing for a unit root in time series regressions,
Biometrika”. 75(2). ss.335-346.
Pindyck, R. S. and Salimano, A. 1993. “Economic instability and Aggregate
investment, National Bureau of Economic Research”, 8, 259-318.
Winters, P. R. 1960. “Forecasting sales by exponentially weighted moving averages,
Management Science”, 6, 324-342.
9