Academy of Economic Studies, Bucharest

Doctoral School of Finance and Banking

Financial Development and Economic Growth

in Romania

Dissertation Paper

Supervisor: MSc Student:

Professor Moisa Altar Mihai Tarau

CONTENTSCONTENTS

1. Introduction

2. Economic Growth-Financial Development Nexus .3. Theoretical Model of Analysis

4. Data Used and Econometric Estimations .

5. Conclusions

1. Introduction

Reasons why countries grow at different rates

The pozitive link between financial intermediation and economic growth

Financial depth and monetization of economy, as exogenous variables for output growth

Simplified mathematical model: ARDL(1,1) & ECM

Econometric estimations: OLS, UVAR & VECM (The case of Romania output growth from January 1992 to September 2001)

2. Economic growth-financial development nexus

Adams (1819) asserted that banks harm the “morality, tranquility, and even wealth” of nations

Hamilton (1781) argued that “banks were the happiest engines that ever were invented” for spurring economic growth

Robert Lucas (1988) dismisses finance as a major determinant of economic growth

Merton Miller (1998): “financial markets contribute to economic growth is a proposition almost too obvious for serious discussion.”

Banks-Markets

Distiction and rivalries

Advantages and Disadvantages

It’s not banks or markets, it is banks and markets; these different components of the financial system ameliorate different information and transaction costs.

Five Key Components (Prerequisites) of a Performant Financial System:

• Sound public finances and public debt management

• Stable monetary arrangements

• Variety and influence of banks

• Credible Central Bank in stabilizing domestic finances and managing international financial relations

• Well-functioning securities market

3. Theoretical Model of Analysis

ARDL (1,1):

(1)

(2)

(3)

Long-run relationship (y stable) requires |b|<1.

(4)

t is distributed independently from εt

ECM from ADL:

(1-b) = speed of adjustment

The long-run (steady-state) relationship implied by the dynamic system in equations (1)-(4) is given by:

(7)

or (8)

The main assumption is that there exist a single long-run relationship between the endogenous and forcing variables. The pre-requisites for consistent and efficient esti-mation are that the shocks in the dynamic specification be serially uncorrelated and that the forcing variables be strictly exogenous.

4. Data used and econometric estimations

The time series used are:

qind = monthly value of industrial production in ROL;rtcrqind = (Δqindt/qindt-1) = industrial production monthly value rate of growth (chain), as a proxy for GDP monthly rate of growth;fd = monthly credit to non-governments in ROL;fdinq = (fd/q) = financial depth (credit to non-governments weight in monthly industrial production);rtcrfd = (Δfdinqt/fdinqt-1) = financial depth monthly rate of growth (chain);m2 = monthly value in ROL of monetary aggregate M2;m2inq = (m2/q) = monetary aggregate M2 weight in monthly industrial production;rtcrm2inq = (Δm2inqt/m2inqt-1) = monthly growth rate of M2 weight in industrial production (chain).

Figure 1 : Time Series Used

0.0E+00

2.0E+07

4.0E+07

6.0E+07

8.0E+07

1.0E+08

1.2E+08

92 93 94 95 96 97 98 99 00 01

FD

1

2

3

4

5

92 93 94 95 96 97 98 99 00 01

FDINQ

0.0E+00

5.0E+07

1.0E+08

1.5E+08

2.0E+08

2.5E+08

92 93 94 95 96 97 98 99 00 01

M2

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

92 93 94 95 96 97 98 99 00 01

M2INQ

0

10000000

20000000

30000000

40000000

50000000

60000000

92 93 94 95 96 97 98 99 00 01

QIND

-0.4

-0.2

0.0

0.2

0.4

0.6

92 93 94 95 96 97 98 99 00 01

RTCRM2INQ

-0.4

-0.2

0.0

0.2

0.4

0.6

92 93 94 95 96 97 98 99 00 01

RTCRQIND

-0.4

-0.2

0.0

0.2

0.4

0.6

92 93 94 95 96 97 98 99 00 01

RTCRFD

Granger Tests Conclusions:Industrial production is the endogenous variable in any relationship (at least under an one year lag) involving credit to non-governments and monetary aggregate M2, as it do not Granger cause any of the last two indicators.

On the other hand, Granger tests reveal an extremely stable causality link between industrial production (as predicted variable) and credit to non-governments or monetary aggregate M2 (as predicting variables) in any lag of at most one year.

Credit to non-governments is, in his turn, Granger caused by the monetary aggregate M2, all over one year period (thus having another time stable causality relationship).

Financial depth rate of growth has a strong capacity of prediction over the industrial production rate of growth, for periods from one month to six months; the same relationship occur between monthly growth rate of M2 weight in industrial production and the industrial production rate of growth, but only for one quarter or one semester lags.

Selected Regression of rtcrqind

Dependent Variable: RTCRQINDMethod: Least SquaresDate: 06/26/02 Time: 16:34Sample(adjusted): 1994:01 2001:09Included observations: 92Excluded observations: 1 after adjusting endpoints

Variable Coefficient Std. Error t-Statistic Prob.

C 0.050523 0.011701 4.317840 0.0000RTCRQIND(-24) -0.357926 0.159371 -2.245867 0.0274

RTCRFD(-1) 0.135798 0.061531 2.206980 0.0301RTCRFD(-3) -0.514537 0.146024 -3.523658 0.0007

RTCRM2INQ(-3) 0.638021 0.150882 4.228618 0.0001RTCRM2INQ(-24) -0.415388 0.153725 -2.702148 0.0084

DUMMAR97 0.488365 0.065471 7.459235 0.0000DUMJAN98 0.455259 0.065044 6.999200 0.0000

DUMJAN2000 -0.331202 0.073907 -4.481318 0.0000

R-squared 0.688115 Mean dependent var 0.040367Adjusted R-squared 0.658054 S.D. dependent var 0.109016S.E. of regression 0.063748 Akaike info criterion -2.575050Sum squared resid 0.337297 Schwarz criterion -2.328353Log likelihood 127.4523 F-statistic 22.89045Durbin-Watson stat 1.856950 Prob(F-statistic) 0.000000

It can be also remarqed that the growth rates of financial depth and M2 weight in industrial production (which explain in the selected regression over 65% of growth rate from industrial production) proved themselfes earlier to be relevant regarding Granger causality tests.

-15

-10

-5

0

5

10

15

00:04 00:07 00:10 01:01 01:04 01:07

CUSUM 5% Significance

Regression

Stability

Test:

Figure 2 : Impulse Responses Functions of rtcrqind, rtcrfd and rtcrm2inq

-0. 04

0. 00

0. 04

0. 08

0. 12

1 2 3 4 5 6 7 8 9 10 11 12

Res p ons e o f RTCRQIND to RTCRQIND

-0. 04

0. 00

0. 04

0. 08

0. 12

1 2 3 4 5 6 7 8 9 10 11 12

Re s p ons e o f RTCRQIND to RTCRFD

-0. 04

0. 00

0. 04

0. 08

0. 12

1 2 3 4 5 6 7 8 9 10 11 12

Res po ns e o f RTCRQIND to RTCRM 2INQ

-0. 12

-0. 08

-0. 04

0. 00

0. 04

0. 08

1 2 3 4 5 6 7 8 9 10 11 12

Re s p ons e o f RTCRFD to RTCRQIND

-0. 12

-0. 08

-0. 04

0. 00

0. 04

0. 08

1 2 3 4 5 6 7 8 9 10 11 12

Res po ns e o f RTCRFD to RTCRFD

-0. 12

-0. 08

-0. 04

0. 00

0. 04

0. 08

1 2 3 4 5 6 7 8 9 10 11 12

Res p ons e o f RTCRFD to RTCRM 2INQ

-0. 12

-0. 08

-0. 04

0. 00

0. 04

0. 08

1 2 3 4 5 6 7 8 9 10 11 12

Res po ns e o f RTCRM 2 INQ to RTCRQIND

-0. 12

-0. 08

-0. 04

0. 00

0. 04

0. 08

1 2 3 4 5 6 7 8 9 10 11 12

Res p ons e o f RTCRM 2INQ to RTCRFD

-0. 12

-0. 08

-0. 04

0. 00

0. 04

0. 08

1 2 3 4 5 6 7 8 9 10 11 12

Re s p ons e o f RTCRM 2INQ to RTCRM 2INQ

Response to One S.D. Innovations ± 2 S.E.

Figure 3 : Variance Decomposition for rtcrqind, rtcrfd and rtcrm2inq

0

20

40

60

80

100

1 2 3 4 5 6 7 8 9 10 11 12

Perc en t RTCRQIND v ari anc e due to RTCRQIND

0

20

40

60

80

100

1 2 3 4 5 6 7 8 9 10 11 12

Pe rc e n t RTCRQIND v a ria nc e d ue to RTCRFD

0

20

40

60

80

100

1 2 3 4 5 6 7 8 9 10 11 12

Perc en t RTCRQIND v a rian c e du e to RTCRM 2INQ

0

20

40

60

80

100

1 2 3 4 5 6 7 8 9 10 11 12

Pe rc e n t RTCRFD v a ri anc e due to RTCRQIND

0

20

40

60

80

100

1 2 3 4 5 6 7 8 9 10 11 12

Perc en t RTCRFD v a ria nc e d ue to RTCRFD

0

20

40

60

80

100

1 2 3 4 5 6 7 8 9 10 11 12

Perc en t RTCRFD v a rian c e du e to RTCRM 2INQ

0

20

40

60

80

100

1 2 3 4 5 6 7 8 9 10 11 12

Perc en t RTCRM 2INQ v a ri anc e due to RTCRQIND

0

20

40

60

80

100

1 2 3 4 5 6 7 8 9 10 11 12

Perc en t RTCRM 2 INQ v a ria nc e d ue to RTCRFD

0

20

40

60

80

100

1 2 3 4 5 6 7 8 9 10 11 12

Pe rc e n t RTCRM 2INQ v a rian c e du e to RTCRM 2INQ

Variance Decompos ition

After proceeding a Johansen test for three I(1) integrated variables (qind, fdinq and m2inq), the cointegrated equation obtained support the initializing of a VEC model.

Figure 4 : Impulse Responses Functions of qind, fdinq and m2inq

-500000

0

500000

1000000

1500000

2000000

2500000

10 20 30 40 50 60 70 80 90 100

Res po ns e o f QIND to QIND

-500000

0

500000

1000000

1500000

2000000

2500000

10 20 30 40 50 60 70 80 90 100

Res p on s e o f QIND to FDINQ

-500000

0

500000

1000000

1500000

2000000

2500000

10 20 30 40 50 60 70 80 90 100

Re s p ons e o f QIND to M 2 INQ

-0. 2

-0. 1

0. 0

0. 1

0. 2

0. 3

10 20 30 40 50 60 70 80 90 100

Res p on s e o f FDINQ to QIND

-0. 2

-0. 1

0. 0

0. 1

0. 2

0. 3

10 20 30 40 50 60 70 80 90 100

Re s po ns e o f FDINQ to FDINQ

-0. 2

-0. 1

0. 0

0. 1

0. 2

0. 3

10 20 30 40 50 60 70 80 90 100

Res po ns e o f FDINQ to M 2 INQ

-0. 3

-0. 2

-0. 1

0. 0

0. 1

0. 2

10 20 30 40 50 60 70 80 90 100

Re s p ons e o f M 2 INQ to QIND

-0. 3

-0. 2

-0. 1

0. 0

0. 1

0. 2

10 20 30 40 50 60 70 80 90 100

Res po ns e o f M 2 INQ to FDINQ

-0. 3

-0. 2

-0. 1

0. 0

0. 1

0. 2

10 20 30 40 50 60 70 80 90 100

Res pon s e o f M 2 INQ to M 2 INQ

Response to One S.D. Innovations

Figure 5 : Variance Decomposition for qind, fdinq and m2inq

0

20

40

60

80

100

10 20 30 40 50 60 70 80 90 100

Perc e nt QIND v a ria nc e d ue to QIND

0

20

40

60

80

100

10 20 30 40 50 60 70 80 90 100

Perc en t QIND v a rian c e du e to FDINQ

0

20

40

60

80

100

10 20 30 40 50 60 70 80 90 100

Pe rc e nt QIND v a ri an c e due to M 2 INQ

0

20

40

60

80

100

10 20 30 40 50 60 70 80 90 100

Perc en t FDINQ v a ria nc e d ue to QIND

0

20

40

60

80

100

10 20 30 40 50 60 70 80 90 100

Pe rc e nt FDINQ v a rian c e du e to FDINQ

0

20

40

60

80

100

10 20 30 40 50 60 70 80 90 100

Perc en t FDINQ v a ri an c e due to M 2INQ

0

20

40

60

80

100

10 20 30 40 50 60 70 80 90 100

Pe rc e nt M 2 INQ v a ria nc e d ue to QIND

0

20

40

60

80

100

10 20 30 40 50 60 70 80 90 100

Perc en t M 2 INQ v a rian c e du e to FDINQ

0

20

40

60

80

100

10 20 30 40 50 60 70 80 90 100

Perc en t M 2 INQ v a ri an c e due to M 2INQ

Variance Decomposition

5. Conclusions Long-run stability of the dynamic relationship between output growth and financial deepening (observed like weighted non-governmental credit and monetary aggregate M2 by the industrial production value)

The limited effectiveness of the banking system is revealed by the extremely low level of banking sector credit in the economy, as banks are unable or unwilling to lend (the last possibility is best mirrored by the crowding-out effect that had a peak in 1999).

Although the monetary aggregate M2 weight in industrial production seems to Granger cause the other variables and to be exogenous (indicating policy effectiveness), the responses of economic growth (and even financial depth) to shocks in its level

are counterintuitive.

The model may be augmented in case of finding better “proxy” variables for economic growth and financial intermediation, or we add other relevant variables omited (e.g. fiscal variables, bank supervision, asset prices a.s.o.).