Financial sector reform and its impact on investment and economic growth: an econometric approach.

Hasan, M. Aynul ; Khan, Ashfaque H. ; Ali, S. Sajid 等

INTRODUCTION

The financial sector is central to economic development as it serves the role of intermediary by mobilising savings and subsequently allocating credit for productive activities. However, in many developing countries including Pakistan, administered interest rate, domestic credit controls, high reserve requirements, use of captive banking system to finance large budgetary requirements of the government and controls on international capital inflows have remained the main features of the monetary policy. These repressive policies had their repercussions in the form of excess liquidity with the banking system, disintermediation of cash flows, segmentation of financial markets, underdeveloped money and capital markets, etc. [McKinnon (1973) and Shaw (1973)], therefore, argued that low interest rate ceilings unduly restrict the real flow of loanable funds, thus depressing the quantity of productive investment.

Financial liberalisation, on the other hand, is defined as policy measures designed to deregulate certain operations of the financial system and transform its structure with a view to achieving a liberalised market oriented system with an appropriate regulatory framework. The financial sector reforms would lead to increase in loanable funds by attracting more household savings to bank deposits due to higher interest rates. This, in turn, would result in greater investment and faster economic growth.

In Pakistan, various measures have been undertaken in the early 1990s to liberalise the financial sector as part of the overall structural adjustment programme (SAP) with the objective to improve the effectiveness of monetary policy. These policies were implemented by making a shift from direct to indirect monetary control and greater reliance on market forces. The main financial liberalisation policies were aimed at liberalising interest rates, reducing controls on credit, enhancing competition and efficiency in the financial system, strengthening the supervisory framework, promoting growth and deepening of the financial markets. In this context, the following measures have already been implemented to date as part of the broader financial sector reforms:

* Efforts have been made to enhance the health of and competition within the banking sector by privatising two nationalised banks, as well as allowing 11 new scheduled commercial banks to be set up in the private sector.

* Debt management reforms were introduced to promote primary and secondary securities markets. The prudential supervisory framework has been established to foster sound credit decisions.

* Interest rate rationalisation was introduced by paring down concessional and direct credit schemes.

* The exchange and payment reforms have also been undertaken in the areas of foreign investment and foreign trade.

The adoption of these measures would give way to greater flexibility in the interest rate movements, an enhanced role for the market forces in credit allocation, gradual deepening of the money and securities markets, and enhancing competition and efficiency in the financial system. Against this background, the paper develops and estimates a medium-sized 24-equation macroeconometric model for the financial sector of Pakistan. The model provides a detailed consistent treatment of financial variables by disaggregating the financial assets held by the households, private business and enterprises. The behaviour of demand for these assets is then linked with the overall national saving and, subsequently, with investment and gross domestic product (GDP). This model will be useful device in not only generating ex-ante forecasts but, more importantly, it will provide answers to numerous interesting and critical counterfactual policy questions in the context of Pakistan's financial sector reforms. For example, if the reforms had to take place in the early eighties rather than the nineties, the model will quantitatively estimate the counterfactual loss foregone in terms of lower GDP, savings and investments in Pakistan. It is expected that these counterfactual policy simulation results may be useful to the policy-makers in designing more accurate and practical future monetary policies in Pakistan.

II. A MODEL FOR FINANCIAL SECTOR

While the financial sector reform is initiated by bringing about changes within the monetary sector only, the impact of such a reform, however, is expected to be multifacit and wide-ranging influencing many other sectors of the economy. A single equation approach, in this context, to evaluate the implication of this reform may not only be inadequate but, at the same time, it could even be misleading. This is due to the fact that, with a single equation approach, an increase in the interest rate for deposit, for instance, may show a greater demand for time deposits. But in terms of total financial assets, the impact of such a change on it may be ambiguous because of the possibilities of substitution among other financial assets (e.g., currency, unfunded debt, floating debt and permanent debt).

Thus, in order to examine the intra as well as inter sectoral impact of financial sector reforms, we have constructed a medium-sized 24-equation macroeconometric model for Pakistan. The model is not only dynamic and rich in specifications but, more significantly, it is based on a pragmatic approach, which takes into account some of the specific institutional arrangements present in the financial sector of Pakistan. Broadly, the model has been divided into five blocks, namely monetary, savings, government revenue, macroeconomic and definitional blocks. In the following, we present the model without discussion so as to conserve the space.

A. Monetary Block

1. Net Financial Assets

NFA = CC + TD + UFDBT + [bar.FDEBT] + [bar.PDEBT] - L

2. Currency in Circulation

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

3. Total Deposit

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

4. Unfunded Debt

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

5. Demand for Loan

L = f CP, [[bar.R].sub.A])

6. Price Level

[P.sub.g] = f (TD/Y, ITR/Y, [P.sub.m])

7. Net Interest Bearing Financial Assets

TIBFA = NFA - CC

B. Savings Block

8. Savings in Real Assets

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

9. Real Financial Assets

RFS = [DELTA]TIBFA/[P.sub.g]

10. Real National Savings

RNS = RSA + RFS

C. Government Revenue Block

11. Direct Taxes

DTR = f(Y)

12. Indirect Taxes

ITR = f(Y, M)

13. Total Tax Revenue

GTR = DTR + ITR

14. Total Revenue

GR = GTR + [bar.GNTR ] + [[bar.S.sub.UR]]

D. Macroeconomic Block

15. Private Investment

[I.sub.p] = f([bar.RA], TD/Y, [P.sub.m]/P.sub.g], IINFSI)

16. Private Consumption

CP = f[(YR - DTR/[P.sub.g]), [[bar.R.sub.m]], [bar.RD]]

17. Public Investment

[I.sub.g] = f(GR/[P.sub.g], [bar.RFA])

18. Public Consumption

[C.sub.g] = f(GR/[P.sub.g])

19. Import

MR = f (YR, [P.sub.m]/[P.sub.g])

20. National Income

[Y.sub.R] = [C.sub.p] + [I.sub.p] + [C.sub.g] + [I.sub.g] + [bar.[DELTA]AST] + [bar.X] - [M.sub.R]

21. Public Infrastructure

IINF = [I.sub.g] - [I.sub.go]

22. Index of Public Infrastructure

IINFST = [([bar.IINFS] (-1) + IINF)/[[bar.IINFS].sub.0]]

23. National Income at Current Prices

Y = YR x [P.sub.g]/100

24. Imports at Current Prices

M = MR x [P.sub.m]/100

List of Variables

CC = Currency in Circulation

[C.sub.g] = Public Consumption Expenditure

[C.sub.p] = Private Consumption Expenditure

DTR = Direct Taxes

FDEBT= Floating Debt

GNTR = Non-tax Revenue

GTR = Total Tax Revenue

[I.sub.g] = Public Investment

IINESI = Index of Economic Infrastructure Investment

IINF = Public Infrastructure

[IINFS.sub.o] = Initial Stock of Real Public Infrastructure Investment

[I.sub.p] = Private Investment

ITR = Indirect Taxes

L = Loan

M = Imports at Current Price

[M.sub.R] = Real Imports

NFA = Net Financial Assets

PDEBT= Permanent Debt

[P.sup.e.sub.g] = Expected Rate of Inflation

[P.sub.g] = General Price Level

[P.sub.m] = Import Price Index

[R.sub.A] = Interest Rate on Advances

RD = Rate of Return on Time Deposit

RF = Rate of Return on Floating Debt

RFA = Foreign Aid

RFS = Real Financial Assets

[R.sub.m] = Remittances

RNS = Real National Savings

RSA = Savings in Real Assets

RU = Rate of Return on Unfunded Debt

[DELTA]ST = Changes in Stock

SUR = Surcharges

TD = Total Deposits

TIBFA = Net Total Interest Bearing Financial Assets

UFDBT = Unfunded Debt

X = Total Exports of Goods and Services

Y = Nominal GDP

YR = Real GDP.

III. RESULTS

The results are discussed in the following paragraphs. Given the existence of long-run relationship, based on PP cointegration test, we used a simple OLS technique to estimate the behavioural equations of the model. (2) By and large, all the estimated regression equations have high adjusted [R.sup.2] and significant t-values with correct signs implying that the individual equations not only explain the postulated behaviour well in the model but, at the same time, individual stipulated parameters can be meaningfully interpreted. Based on other estimated statistical tests (e.g., LM, ARCH, and CUSUM), we can safely argue that the estimated regression equations are free from the statistical problems of serial correlation, heteroscedasticity and instability of the parameters.

Based on the estimated equations of the model the policy simulation exercise undertaken in this paper is essentially counterfactual in character. The key objective in carrying out this type of exercise is to investigate and unfold the conundrum as to what would have happened in Pakistan, in terms of the impact(measured by key economic variables) and the ensuing monetary cost (or loss of revenue in the form of investment and output), had this country introduced the reforms earlier in the eighties rather than later as done so in the nineties. In order to cover the broader aspects of the financial reform and, at the same time, keeping the discussion more manageable, we report results of the impact of three categories of policy simulation on key selected macroeconomic variables. Furthermore, for each type of policy simulation, we also calculate the corresponding expected loss of revenue (in rupees) due to late implementation of the reforms. The three broad categories of reform considered are:

1. Interest Rate Liberalisation;

2. Spread Reducing Reform; and

3. Financial Deepening.

In the following, we discuss the results of these policy simulations.

1. Interest Rate Liberalisation

As a result of repressive monetary policies pursued in Pakistan, the real deposit rates remained negative most of the time during the decade of the 1980s. This type of policy is expected to make non-bank assets relatively more attractive than the bank deposits thus creating financial disintermediation. In order to test the sensitivity of the above policy reforms, we conduct two types of simulation. Firstly, we keep the nominal deposit rates at least as high as 14 years (1981-94) of average inflation rate. This implies that, on average, the real rate of deposits during this period should be non-zero thus enabling the real return on financial assets to be non-negative. The second type of policy simulation in this regard entails the nominal deposit rates to be at least two percentage point above the 14 years (1981-94) average inflation rate making the real rate positive by two percentage point. The cost of late implementation of the reform is evaluated in terms of loss in real private investment and real GDP. The cost of late implementation in rupee term is reported in Tables 1 and 2.

In the first case when real return on financial assets are assumed non-negative Pakistan's economy could have avoided a loss of almost Rs 12 billion in real GDP on cumulative basis since 1980-81 as shown in Table 1. In fact, if the rate of return on deposits were allowed to be 2 percentage point above the 14 years average inflation rate (implying 2 percent positive real return on deposit), the economy could have saved a loss of Rs 78 billion in real GDP on cumulative basis or Rs 6 billion on average each year since 1980-81 as shown in Table 2.

2. Spread-reducing Reform

Another key impediment to the financial sector reform was the government policy of maintaining a large spread between the deposits and the lending rates of the banking system. Under the spread-reducing financial policy simulation, we have analysed the impact of keeping the differences between the deposit and lending rates at 2 percent on private investment and real GDP. The results, as reported in Table 3, suggest that Pakistan could have saved Rs 22.7 billion in term of loss in real GDP on cumulative basis or Rs 1.65 billion per year since 1980-81.

3. Financial Deepening

Another interesting policy simulation conducted in this study pertains to the impact of financial deepening on the economy. Financial deepening in this context implies a broadening of the monetary base in relation to the real sector of the economy. The basic premise underlying such a reform is that making the financial sector wider and covering larger sectors of the economy will facilitate economic activity and may improve physical investments as well as total output of the economy. This simulation was implemented by increasing the financial deepening variable (TD/Y) by 25 percent as compared to its actual value. Promoting financial deepening policies in the early eighties rather than in the nineties could have saved Pakistan's economy up to about Rs 260 billion in terms of loss in real GDP on cumulative basis or Rs 16.6 billion per year on average basis since 1980-81 (see Table 4).

IV. CONCLUDING REMARKS

While the significance of financial reforms in Pakistan introduced in the early 1990s cannot be undermined in that it has brought about some real changes in terms of freeing interest rates, reducing the spread rates between the deposit and lending rates, privatisation of nationalised banks, and many more, what is, however, important from the public policy point of view in this context is to know how much Pakistan's economy could have saved or gained had the reform been introduced earlier in the 1980s. Analysis of this nature which is also known as counterfactual exercise is useful not only in terms of getting an estimate of foregone benefits of delaying reforms in the past but, more importantly, it provides the policy-makers with a better insight into successfully implementing the reform in the future.

With this perspective in mind, this study develops and estimates a 24-equation medium-sized macroeconometric model for Pakistan with a specific focus on the financial sector. Although the model constructed does not explicitly deal with specific institutional arrangements (privatisation, NBFI, etc.) and management side (prudential regulations, competitions, etc.), of financial reforms, analysis of this paper, nevertheless, quantifies some of the important implications of the reform within the context of policy simulation exercise. Three important areas of financial reform where this study has made some modest contribution are the ones that relate to identifying the impact of more flexible market determined interest rates; reducing the spread between the deposit and lending rates; and finally, promoting policies to improve financial deepening in the economy.

In general, our findings suggest that the impact of all three financial sector reform policies not only reduce financial disintermediation (McKinnon-Shaw Hypothesis) but the positive influence also permeate into the real sector. In fact, the key finding of this study is that had Pakistan introduced the financial sector reforms in the eighties rather than in the nineties, the economy could have enhanced its output in veal terms by over Rs 16.5 billion every year. This figure is by no means a trivial amount given the size of the average annual real output of Rs 344 billion during this period (1981-94).

REFERENCES

Hasan, M. Aynul, Ashfaque H. Khan and S. Sajid Ali (1996) Financial Sector Reform and its Impact on Investment and Economic Growth: An Econometric Approach. The Pakistan Development Review 35:4.

McKinnon, R. I. (1973) Money and Capital in Economic Development. New York: Oxford University Press.

Shaw, E. S. (1973) Financial Deepening in Economic Development. New York: Oxford University Press.

(1) Detailed discussion can be found in Hasan et al. (1996)

(2) The estimated regression coefficients of the model and various test statistics are available on request from the authors.

M. Aynul Hasan is Professor of Economics at Acadia University, Canada. Ashfaque H. Khan is Joint Director at the Pakistan Institute of Development Economics, Islamabad. S. Sajid Ali is Economist at the State Bank of Pakistan, Karachi.

Table 1
Simulation Results of Interest Rate Liberalisation
(Real Return on Deposit in Non-negative)
(Rs in Million)

 Real Private Investment Real GDP
Year (Cumulative) (Cumulative)

1980-81 223 2748
1981-82 418 4937
1982-83 368 4171
1983-84 560 6878
1984-85 492 5742
1985-86 417 4592
1986-87 363 3773
1987-88 640 8457
1988-89 867 12869
1989-90 742 9936
1990-91 1076 15534
1991-92 1060 15066
1992-93 941 13152
1993-94 840 11749
Cumulative Average 106 1341

Note: Cumulative values are simply the sum of
successive marginal values of each year.

Table 2
Simulation Results of interest Rate Liberalisation
(Real Return on Deposit is 2 Percent)
(Rs in Million)

 Real Private Investment Real GDP
Year (Cumulative) (Cumulative)

1980-81 709 8215
1981-82 1339 14951
1982-83 1642 18303
1983-84 2232 26038
1984-85 2467 28835
1985-86 2703 32076
1986-87 2910 35058
1987-88 3434 43364
1988-89 3886 51826
1989-90 3923 52039
1990-91 4617 63100
1991-92 5044 68446
1992-93 5406 73345
1993-94 5746 77743
Cumulative Average 483 5960

Note: Cumulative values are simply the sum of
successive marginal values of each year.

Table 3
Simulation Results of Spread Reducing Reforms
(Difference between the Average Lending and Deposit Rate is 2 Percent)
(Rs in Million)

 Real Private Investment Real GDP
Year (Cumulative) (Cumulative)

1980-81 238 1849
1981-82 496 3784
1982-83 727 5395
1983-84 953 7165
1984-85 1178 8843
1985-86 1381 10333
1986-87 1541 11179
1987-88 1719 12465
1988-89 1860 13209
1989-90 2010 14156
1990-91 2237 16306
1991-92 2466 18074
1992-93 2723 20358
1993-94 2992 22697
Cumulative Average 222 1645

Note: Cumulative values are simply the sum of
successive marginal values of each year.

Table 4
Simulation Results of Financial Deepening
(Financial Deepening Variable Increased by 25 Percent)
(Rs in Million)

 Real Private Investment Real GDP
Year (Cumulative) (Cumulative)

1980-81 1652 17482
1981-82 3153 32283
1982-83 4237 43720
1983-84 5795 62451
1984-85 6867 74999
1985-86 8081 90955
1986-87 9256 106965
1987-88 10848 129702
1988-89 12306 154636
1989-90 13156 169028
1990-91 15092 196951
1991-92 16830 217529
1992-93 18553 239716
1993-94 20262 260489
Cumulative Average 1416 16616

Note: Cumulative values are simply the sum of
successive marginal values of each year.