Policy, Research, and External Affairs Debt and International Finance International Economics Department The World Bank July 1991 WPS 717 Does Fina^ncial Liberalization Really Improve Private Investment in Developing Countries? Jacques Morisset An increase in real interest rates, which is a typical element of financial reforms, does not necessarily involve a positive effect on private investment unless the authorities are careful to ensure that (1) bank deposits are closer substitutes to unproductive assets (cash, gold) and foreign assets than to capital goods, (2) the financial sector assures an efficienlt allocation of domestic credit, and (3) the flow of domestic credit to.the private sector is not absorbed by the needs of the public sector. The Pohcy. Research, and Extemal Affairs Complex disutithues PRI- Working lapers to dissem inate the findtngs 0f work in progress and to encourage the exchange of ideas among Bank staff and all others interested in development issues. 'Ihcse papers carTy the names of the authors, reflect only their views, and should be used and ciled accordingly. The findings, interprctations, and conclusions are the authors' own. They should not be attributed to the World llank, its leard of Directors, its managaecnt, or any of its member counuries. PolOcy, Research, and External Affairsi De L ad .ernallonal Finance WPS '717 This paper - a' product of the Debt and Intemational Finance Division, International Economics Department --is part of a larger cffort in PRE to determine the interaction between external and domestic finance in support of investment in developing countries. Copies are available free from the World Bank, 1818 H Street NW, Washington DC 20433. Please contact Sheilah King-Watson, room S8-045, extension 31047 (22 pages). Assuming that li4uidity constraints exist in most bank deposits reduces the private sector's developing countries, the majority of analysts willingness to hold government bonds, so the believe that increasing real interest rates will public sector must finance a given budget deficit raise the volume of lending and hence private with more domestic credit. investment. His simulations for Argentina for 1961-82 Morisset, focusing on the demand for capital suggest that the low response of private investors goods, argues that the positive effect on the to changes in interest rate policy in those 20 domestic credit* market may be offset by the years was attributable not to the low values of negative effect of a portfolio shift from capital interest elasticities but to the interaction of the goods and public bonds into monetary assets. mechanisms allowed for in the model, which He also demonstrates that a policy of financial tends to neutralize the impact of such policies. liberalization could increase the public sector's demand for domestic credit, thus limiting the Morisset concludes that the effect of changes funds available to the private sector. This in interest rate policy on the demand for capital crowding out does not result from a change in goods is weak in Argentina - and might affect the government's behavior but from a shift in the the quality of private investment more than its portfolio of private agents. Higher demand for quantity. The PRE Working Paper Series disseminates the findings of work under way in the Bank's Policy, Research, and External Affairs Complex. An objective of thc series is to getthese findings out quickly, even if presentations are less than fully polished. The findings, interpretations, and conclusions in these papers do not necessarily represent official Bank policy. Produced by the PRE Dissemination Center Table of Contents Introduction 2 1. A Simple Portfolio Model 3 2. The Flexible Accelerator Model of Investment 8 3. The Introduction of the Public Sector 10 4. An Empirical Test : the Case of Argentina (1961-82) 13 S. Concluding Remarks 16 Bibliogra:phy 17 Tables 19 I am grateful to M. Blejer, A. Fishlow, J. de Melo and two anonymous referees for helpful comments and suggestions. 3 Inteoduction In the economlc literature, the relationship between real intereat and private investment in LDCs has received considerable attention. Until the early 1970s, the economists assumed that low interest rates would promote investment spending and economic growth in accordance with Keynesi&n and neoclassical theories. McKinnon (1973) and Shaw (1973) were the first to challenge seriously this conventional wisdom. These two authors suggested that higher real interest rates would raise savings, increase the volume of domestic credits extended by the financial system and hence the equilibrium rate of investment. In order to assess tho McKinnon-Shaw hypothesis, an impressive number of empirical studies have been carried out during recent year2/. In the present state of research, the tendency is to admit the validity of this hypothesis. As a matter of fact, establishing high real interest rates has become a standard part of the policy advice given to LDC's by external experts, ranging from the visiting academic economist via the World Bank to emissaries of the IH(see Van Wijnbergen (1983) or Polak (1989)). The most favored justification for a high interest rates policy in LDCs is derived from the presence of liquidity constraints on private investment decisions. Policies that impose artificially low interest rate ceilings tend to constrain the supply of capital and lead to an excess demand for capital relative to what would happen if the deposit interest rate were allowed to find its market-clearing level. Because the principal constraint on investment is the quantity, rather than the cost, of financial resources, a rise in interest rates will increase the supply of credit to finance private investment. "Any effect exerted by the rate of interest on private investment is not direct within thls rationing framework but, rather, occurs via the channel of financial savinsl (Blejer an Khan (1984), p.386). Even if a clear consensus has emerged in recent years that a significant fraction of the population in developing countries is affected by liquidity constraints3/, the problematic issue is how far to go. We argue, by focusing on the demand for capital goods, that a number of factors might influence the relationship between real interest rates, the supply of domestic credits and private investment. The purpose of this paper is to confront the HcKinnon-Shaw hypothesis with a simple model of investment behavior. The resulting model is esetimted and simulated for Argentina over the 1° .-82 period, giventhat this country has been affected by various inte.est rates 2/For a review of this literature, see Fry (1988). 3/Note that the rationing in LDCs is justified as a disequilibrium phenomenon caused by legal ceilings on interest rates. By contrast, for developed countries, the argument is based on modern theories of imperfect information. 4 policies during the last twenty years4/. Simulation results indicate that, in the case of Argentina, the quantity of private investment is little responsive to movements in interest rates. While this finding may not be entirely unexpected (e.g. Khatkhate (1988)), it appears that this insensitivity is'not due to the low values of some important parameters, but rather to the interactions of many opposing forces allowed by the model. In particular, we will demonstrate that the positive effect suggested by McKinnon and Shaw may be offset sLultanseously by a crowding out effect and a shift that could occur iu the portfolio of the private sector. The paper proceeds as follows. In section 1 we present a simplified portfolio model. Although this approach is derived from a Tobin-Sidrauski framework, a liquidity constraint is Introduced into the model in accordance with the McKinnon-Shaw hypothesis. The demand for capital accumulation by the users of capital goods is specified in section 2 using a modified version of the flexible accelerator theory of investment. Section 3 is devoted to the demonstration that the positive Impact of a rise in interest rates on the supply of domestic credlts to finance private investment might be reduced or even reversed when the public sector is introduced Into the analysis. The equations constituting the complete model are presented in Table 2. The empirical results for Argentina are presented and discussed in section 4. Finally, section 5 contains our conclusion. 1. A Siple Portfolio Model An increase in real interest rates following a financial liberalization generates, in general, a portfolio shift in favor of bank deposits. The authors in favor of the HcKinnon-Shaw hypothesis assume that this portfolio shift is coming out from unproductive assets such as cash and gold. This seems, however, a drastic simplification of the reality because it is not at all obvious that deposits are closer substitutes to these unproductive assets than to capital goods in LDCs. To illustrate this point, private Investment in LDCs is mostly the demand for capital accumulation by the owner of capital (see Khatkhate (1988, p.580)). This form of savings tends to decline when real interest rates rise and correspondingly saving in financial assets goes up. Unless the latter effect dominates significantly the former effect, private real investment may not register an upwards change with the rise in the real interest rate. In this section, we address the critical issue of whether the positive effect suggested by HcKinnon and Shaw is strong enough to offset the shift from capital goods into mon.tary assets that could occur in the portfolio of private agents. A/Specification using cross-section data from LDCs, while providing many more observations, assumes aa similarity of behaviors which is questionable. As a starting point, we consider the following real budget conseraint for the private sector5/ S (1) apt 4' OLptJP + OcD1P - + Oht 4+ Mo The private sector can accumulate assets into three components s real domestic money ( m), real domestic assets (Oh) and real foreign assets ( J). Eqution (1) states that prlvate expenditure for net accumulatLion of assets is constrained by the amount of real private savings (se), the changes in net real domestlc credits to the prlvate sector extended by the banking system ( L0/P) and the changes in gross real prlvate external debt ( Dp/P). The demand for real monetary assets is defined as the changes in the stock of money H3. Although forelga assets purchasing by residents are not recent, no such dlrect measures are avallable for developing countries. Therefore capital flight is used as representative of thLs form of savings. The definltion proposed by the World Bank (1985) appears to be the most approprlate to measure foreign assets holdligs by the private sector slnce lt does not dlstingulsh between normal' capital flows and capital flights (2) jt,, - (ODt + I*t - R,)/P - ft Thls deflnltion of capltal flight takes inflows of capltal In the form of increases in gross external debt ( D) and net forelgn direct investment (1*) and subtracts from these inflows the current account deficit (f) and the increase in official reserves ( R). The dlfference between these lnflows and the extent to whlch they are used to finance the current account deflclt and an increase in reserves is taken to reflect an increase in net forelgn claims by the private sector ( j). Flnally, real domestic assets (Oh) are nothing but the difference between the financlng avallable to the private sector and the two components defined above. Thuss (3) Qht - Spt + QDpt/P *s- OLp /P - 64 - iet + bbt Real domestic assets include the amounts of physical capital (lp) and public bonds ( b) that the private sector can accumulate. Goverrment bonds are assumed to be net wealth to the private sector because liquldlty constralnts, by preventing the optimal consumption-savings declslons from being realized, can make present taxatlon less desirable to households than future taxatlon (e.g. Haque (1988)). 5/SLnce the model is designed explicltly for empirical testing, we do not distingulsh firms and households ln (1). Indeed, lf we assume that all firms equlties are held by households, this asset can be eliuinated from (1). Furthermore, Tybout (1986) suggested that both behaviors can be speclfied by a portfolio model. The demand for Om, Oh, and Oj can be specified by a standard portfolio model6/. For simplicity, we assume that the amount of private savings is predetermined so that we can write7/s (4a) &a c + is 11y + 4s ' + ss + +4 ) Ra(w/P) / (Cb) Oh - 420 + colY + got + £14(r*+x) + R,Q(ODO/P) + ffWWL,IP) + cnu7p-l (4c) Oj- a + isly + 4 4 + r 4'+ £&.(*+x) + 9N(W/P) + dcA,(P/) + where r is defined as the ex ante domestic real interest rate on deposits, r* as the foreign real interest rate, x as the expected rate of depreciation of the local currency, y as real income, wo as the expected rate of inflation and wp as net real wealth accumulated by the private sector. The demand functions (4) are based on the theoretical arguments proposed by Brainard-Tobin (1968) and Purvis (1976). These authors suggested that assets demands depend essentially of the disposable income, the level of private wealth held in beginning of period and the rates of return on alternative assets. We defined the return on money as the ex ante real interest rate on deposits and admitted that the demand for domestic assets is positively correlated with the expected rate of inflation. The expected rate of inflation enters to represent portfolio shifts towards capital goods and government bonds (indexed assets) as the expected real rate of return on money falls (see Tobin (1965), Sidrauski (1967) and Fisher (1979)). Note that we do not distinguish the demand for capital goods and government bonds because the rates of return on both assets are generally indexed in developing countries with high and variable inflation. Further desegregation of domestic assets will take place in section 2. We defined the expected Late of return on foreign assets as the foreign real interest rate and the expected rate of depreciation of the local currency. In order to take into account the existence of liquidity constraints on portfolio's decisions, the variations in bank credits to the private sector ( Lp) and in foreign capital iflows ( Dp) have been introduced into the model. An increase in real credit or in foreign financing will in general encourage the acquisition of assets since the economic agents are limited by the availability of their current resources. The expected signs for the parameters of equation (4) are sumrized in table 1. Each rate of return influences positively the 6/A similar model has been presented in more details in Morisset (1989). 7/The decision between present consumption and future consumption does not seem to be significantly influenced by the real rate of interest in most developing countries (e.g. Giovannini (1983) or Khatkhate (1988)). S - ~~~~~~~~~~~~~~7 asaet to which it is associated and negatively others assets8/. We also assume that an increase in GDP level raises the demand for money (transaction motive) and decreases the demand for foreign assets (see Conesa (1986)). In the presence of borrowing constraints, one increase in L. or In OD. is expccted to be positive on the three assets demands. Recently. va.ious authrs argued In favor of a positive relationship between the variation In gross extemal debt and capital flight (e.g. Cuddington ^0987)) and Lessard and Williamson (1987)). In view of the budget constraint (1) it Is clear that the three asset demand functions are linearly dependant. Once an agent has determined his holdings of any two assets given the level of the rescaurces available to him, his demand for the third asset has implicitly been determined as well. Thus the following restrictions must holds (5S-) (Il + E8J+ 4S - 0 with j - 1,2,3,4 and 7 ON ( tJ + aj 4+ gJ a I vwith j - 5 and 6 Hence equations (4) are linearly dependant and one equation must be omitted for the estimation of the model. The model (1)-(4) is similar to the one presented by Tobin (1965) in the sense that an increase in the demand for financial savings following high real interest rates on deposits will generally lead to a reduction in domestic assets demands. ThLe portfolio shift represents the larger attractiveness of holding money than productive capital and public bonds. However, this approach fails to take into account the McKinnon-Shaw hypothesis which assumes a positive relationship between the demand for money and the demand for capital accumulation via the domestic credits to the private sector extended by the banking system. This positive link between money demand and real investment may be easily introduced into the model through the real budget constraint of the financial sector. The banking system, which is assumed to integrate. the central bank and cow-arcial banks, accumulates reserves, extends credit to the government and the private sector, and issues liabilities in the form of money: (6) (6LP/P) = m - (6R/P) (OL6IP) + nw where the real banking system net profit (Onw) Le defined as the difference between receipts and outlays, i.e. Interest received minus interest paid plus other net non-interest Income minus operating expenses, corrected by the changes in domestic inflation and exchange rates9/. 8/In other words these three assees are gross oubstitutes. 9/We can express the banking system net profit as s Equation (6) indicates that a rise in money demand must cause, ceteris paribus, an Increase in the supply of credit to the private sector since domestic credit is the primary asset backing the monetary liabilities of the banking system. Moreover, this presentation also emphasizes that the amount of credit to the private sector is not directly controlled by the financial sector In most developing countries. First, controls on credit to the private sector have been the principal lnstruments of the monetary policy and, second, the amount of credit to the public sector has been usually determined by the demand of the government rather than the supply of the financial systemlO/. Introducing (6) into the system (l)-(4), we can reconcile the Tobin's arguments and the HcKinnon-Shaw hypothesis. In a financial liberalization program, the real rate of interest on deposits and lending rates are In general both decontrolled. The variation in the deposit interest rate should encourage financial savings accumulation while the variation in lending rates should assure positive profits by the banking system. Note, however, that the variation in lending rates does not affect directly assets demands because, under rationing, the cost of credit is not an argument of the private sector's portfolio behavior. The total impact of a variation in real interest rates on the demand for domestic assets isll/ & d(Ah) ana 7) - 3 + or < O dr (1 - ld The first term corresponds to the substitution effect suggested by Tobin and the second term to the positive effect postulated by McKinnon and Shaw. A variation (say an increase) in the real interest rate changes the asset portfolio as assetholders move out of real savings and foreign assets into domestic money. On the other hand, the rise In money demand increases the supply of domestic credits to the private sector and private investment since the private sector is assumed to be liquidity constrained. The total impact of a change in the interest rate Om - (i Lp -)L p-F - (its - +) (i* - - z)R_.1P - (i0 -) -t + ni where IL. is defined as the nominal lending interest rate to the private sector, iL. as the nominal lending interest rate to the public sector, 1* as the foreign nominal interest rate, L. as the nominal interest rate on deposits, ni as other net non-interest income, i' as the inflation rate, P as price level and z as the rate of depreciation of the local currency. 10/see section 4 for a more precise discussion on this remark. lI/For simplicity, we assume that real disposable income and financial sector9s net profits are not influenced by changes in real interest rates In equation (7). Higher borrowing and lending rates should affect these two variables in opposite directions so that the total effect should be weak. 9 depends oan whether the increase in domestic credites due to the McEinnon- Shaw effect exceeds or not the reduction ln the demand for capital goods due to to the portfolio shLft. Note that this approach does not take Into account the eventual declcIe in ezcesslve inventorLes when firms can Invest in an alternative domestic fLnancLal assets. If inventorLes are reduced to normal level, productive investment may increace. The upshot of all thes is that ono camot predict the implication of changes in real deposits rates without having some Lnsight in the financial structure of the economy. Specifically, one wants to know which asset is the closest substLtute to ban depoolts (i.e the coeffLcients oil. £2, and implicitly £,) and whether the banklng system wlll be allcwed to play lts intermediary role or not (L.e the coefficients ,,., C. and implicLtly C..). In that sense, a financial liberalizatLon program may dLrectly improve the allocation of domestic credit to the extent that, with positlve real interest rates, credlt would be allocated according to expected productivLty rather than transactLon costs and percelved risks of default (see M4Kenlon (1973)). Since financLal liberalization policies may lnvolve changes not only ln the quantLty of lnvestment but also in the quality, we wlll dicuss more precisely on this point In the empirical part of the paper (see sectLon 4), 2. The Flexible Accelerator Model of Investment The approach presented in section 1 clearly stresses the substLtutLon effects generated by changes in interest rates on assets markets. The portfolio model owes much to the pioneering work of Van Vijnbergen (1983) even if the portfolio shift into bank deposits Ls coming out from real savings instead of the "curb marketg as suggested by this author. Because the majority of the authors who attempt to study the relationship between real interest rates and real private investment in LDCs has viewed investment only as the demand for capital goods by the users of capital servLces, we introduce into the analysis this second aspect (e.g. Blejer and Khan (1984), Pry (1980) and Tun Wai and Wong (1982)). The demand for capital servLces Ls usually derived from an adapted version of the flexible accelerator model. This approach emphasizes the effects of the resources constraints faced by private investors in accordance wLth the McKinnon-Shaw view. In the long-run representation of the accelerator model, the desired stock of capital (k*W) that the private sector wishes to have in place can be assumed to be proportional to expected output (y*) : (8) k*P Sy* This is a quite standard formulation in which the underlying production function has (technologically) fixed proportions among 10 factorS inpUts so that factor priceS do not enter into the specificationl2a/ it iLs asumed that there is only a partlal adjustment of the privte'ts sector actual capltal stock to lto deolred level so that we can wrlte I (9) bkp m B(k*0 k ,) where k, is the actual private capital stock, Olt, Ls not private lnvestment, and B is the coefficient of adjustment. Gross private lavestment is made up of two components - net inveotment and replacement. Net Investment results from changes In the desired stock of capital, whlle replacement is here assumed always to be a fractLon a of the capital stock on hand at the boginning of the period s (10) ip - Okp + akp X We then substltute equations (8) and (9) lnto equatLon (10), and we get the following Investment functlon for the private sectors (11) Lp - Bey* - ( - d)P, In order to take lnto account the exeistence of llquidlty consetraints, the speed of adjustment between desired and actual private capltal stock (B) is assumed to vary systematlcally with the availabllity of internal funds, bank credits and foreign capital inflows. A linear representation of this relationshlp is s (B,cf + 132(OL,/P) + 138(6D,/P) (12) a 1 so + with B3 > 0 i 1,2 and 3 (k*o - k,-l) where cf is defined as flrms' cash flow; a measure of lnternal funds available for financing private investment. Equation (12) states that the speed of adjustment is influenced posltively by the total flnancing available to the private sector measured in relative terms with respect to the size of the discrepancy bet-een desired and actual private capital stock. Note that if the signs of the parameters in equation (12) are all expected to be posltive, the impact of the three alternative sources of financing on the speed of adjustment is not assumed to ba equal. As bank credlts depend on the flow supply extended by the financial sector rather than the demand of private investors, a flrm's investment decision is independent of its financial condition. Recently, Fazzarl, Hubbard and Petersen (1988) demonstrated that lnternal and ,ternal resources are not perfect substitutes in a firm with liqu4ity constraints. 12/ Using an alternative production function, the desired capital stock level could be also influenced negatively by the rental price of capital. While this specification may complicated the empirical implantation of the model - such variable cannot be easily calculated for developing countries, it does not change significantly the results presented here. Finally, subsetituting equation (12) into equation (11), we obtain the following private investment function t (13) ip oly, + /41cf + P1S(tQL,/P) + pI6(D /P ) + plap_1 with1sc o 30 a _ G0 PM 8 0 A riso in the reaources available to the private sector increases the speed of adjustment so that all parametere in equation (13) could a priori be determined positive. If we incorporate this equation into the portfolio model presented in sectinn 2, the effect of an increase in real interest rates on private investment appeare to be unambiguously positivel3/ s dip Piass (14) - - > 0 dr (l 1a,) The general conclusion is that an increase in the real rate of Interest is favorable to financial savings which, in turn, Increases the amount of bank credits available to the private sector. This rise influences positively the speed of adjustment between desired and actual private capital stock and, hence, private investment. But this presentation provides a misleading and inconclusive picture of effectiveness of interest policies. In particular, the introduction of the public sector into the analysis may affect dramatically the former result since the government can absorb part of the resources made available for private investment. The next section is devoted to the demonstration of this crowding out effect. 30 The Introduction of the Public Sector The experience of the last decade Indicates that most LDC government financed their fiscal deficit with credits from the central bank. Although these governments could conceivably experiment with this source of financing if it were an independently controllable variable, it would generally be more realistic to treat additional borrowing from the central bank "0as a consequence of fiscal disequilibrium or of the inability or unwillingness to finance the growth of exhaustive government expenditures,subsidies, and transfers through explicit taxes or the sale of government debt to the private aectorB (von Purstenberg (1983, p.233)). Following this remark, the demand of the public sector for central bank credits can be specified as s (15) OLO/P ° def0 - Ob - 6D9/P 13/Without loss of generality, we assume that the cash flow of firms is not Influenced by a variation in borrowing and leading interest rates because the impact of such variation is difficult to prodict. 12 where def0 ls defined as the total fiscal deficitl4/, Ob as the sales of public bonds to the private sector, L /P as the changes in domestic credits to the government extended by the banking system and D /P as the variations In gross public external debt. Equation (15) empLaizes that most LDC goverment have financed theig deficlt with central bank credits when they cannot use other sources of financing. Indeed, in many developing countries financing the pub1Lc deficit in the foreign and domestic credit markets becomes less feasible over time because the declining credibility of the public sector. The authorlties can administer the bonds war' at by two different instrumentss interest rates ceilings and controls on the flow supply of bonds. If either the interest rate on public bonds or the flow supply of bonds is treated as an exogenous varlable, the other become eadogenous. With the interest rate exogenous, the goverment chooses to fix legal ceilings on the lnterest rate on public bonds. In theis case, the quantity of bonds is only determined by the bonds demand of the private sector, Alternatively, the government can choose the amount of bonds exogenously and the public bonds market adjusts through a variation in the interest rate on bonds. Let us assume that the interest rate on public bonds is fixed and that the quantity of public bonds is determined by the private sector's willingness to hold public bonds. From equation (3), the demand of the private sector for public bonds is equal to : (16) Ohb - Oh - Le The presence of the public sector can alter tne positive effect of financial liberalization on domestic credlts and, hence on private Investment. Introducing equations (15) and (16) lnto the model, the total impact of a variation in real interest rates on the quantity of bank credits available to the private sector is the followingUl/ 14/The total fiscal deficit can be expressed as i def - [(Lo + g - t] 4 (Lb - f)B3/P + (It - W)LO,1/P +(i* - W /P where ib is the nominal interest rate on public bonds, 1i the nominal lending rate on domestic credit to the pubilc sector, i the foreign nominal interest rate, l. public investment, g public current expenditures, t net taxes, B.. the nominal stock of pub1Lc bonds in the beginning of perLod, L,-1 the nominal stock of domestic eredLt to the pubLic sector, D.-, the nominal stock of public external debt and P price level. 15/AgaLn, for simplification, we assume that a variation in interest rates does not affect the public deficLt. In fact, even If the interest rate on public bonds Ls assumed to be fixed, the pubLc debt-service payments will increase through the change in lending rates so that the crowding out of domestic credit to the private sector is stronger than suggested in (17). 13 d(6Lg,,P) (gas, + ua2) (01 + 2) (17)- - > or < 0 dr ( l I e o (2 + pw) 089 + s) The total impact of an increase in the real interest rate on bank credits available to the private sector cannot be a priori determined in multiplier (17). As suggested by McKinnon and Shaw, a rise in the real rate of interest increases the demand for national liquidities and, therefore, the quantity of bank credits. But, the introduction of the public sector into the analysis sheds light on an important channel through which crowding out of domestic credits to the private sector occurs in many developing countries. The financial reform may exacerbate the demand of the public sector for bank credits and, thereby, limlting the funds available to the private sector. It is noteworthy that the specification chosen here suggests that the crowding out does not come from a change in the government's behavior, but from a shift in the portfolio of private agents. As demands for money and capital goods increase, the private sector's willingness to purchase government bonds is reduced, constraining the public sector to finance its deficit with more credits from the central bankl6/. Interestingly, the crowding out effect does not appear very sensitive to changes in different sources of financing of the fiscal deficit. In order to prevent the diversion of domestic credit to the public sector following an increase in interest rates on bank deposits, the government may be attracted to increase simultaneously the interest rate on public bonds or, if the interest is endogenous, to raise the flow supply of bonds. In this case, the crowding out of domestic credit to the private sector will not come from the portfolio shift from domestic bonds into monetary assets, but from the increase in the public debt service on domestic debtl7l. Similarly, the authorities may attempt to increase the gate of inflation an usual source of financing in a large number of LDCs° economios. The impact of inflationay finance on bank credit remains amblguous. On one hand, the effect is positive through the increase in public bonds demand (indexed assets) and the reduction in the public debt-service but, on the other hand, the effect is negative through the decline in money demandl8/. 16/ The sam argument was used a decade ago by various authors in order to justify Interest controls in LDCs (e.g. Nicole (1974) and Polak (1989)). These authors suggested that lf the rate of interest on bank deposits is kept low, investors lack an attractive alternative to investing in government paper. 17/An increase in the interest rate on public bonds raises the debt service of the domestic debt and, therefore, the public deficit. 18/Of course, this discussion on the effect of a variation in the rate of inflation is tentative because inflation is treated as an exogenous Variable In the model and the Oliveira-Tansi effect Is not taken into account. 14 Using equation (17), we can successively define the short=inm total effect of a change in real interest rates on the demand for domeotic assets and private investment s dOh 3e(la + (18) - a L8+ or 6 0 dc M98 + Ihis) dip P/S(W + (2a) (19) -c . _or < O dr (0 + PS-) Equations (18) and (19) suggest that the effect of financial liberalization cannot be a priori determined on domestic assets demand and private investment. In summary, the complete model consist of three behavioral relationships - equation (4a), (4b) and (13) - and four technical relationships - equations (1), (6), (15) and (16). Together these seven equations can be used to explain AM, OH, 07, OB, OLp and OL.. The complete model is presented in Table 2. Within this framework, we address the critical issue of whether the positive effect suggested by McKinnon and Shaw is strong enough to offset the crowding out of domestic credits and the shift from real savings into monetary assets that could occur in the portfolio of the private sector. 4. An EMgirical Test s the Case of Argentina (1961-82) The model developed in the last section could be used in principle to examine the short-run effect of a variety of shocks on private investment and other endogenous variables of the model, By estimating the model for Argentina and by stimlating it, the response of key variables to changes in real rates of interest will be derived for this country. The data on S., Y, ON, F, I*. and AR have been obtained from the -International Monetary fund's International Financial Statistics. Series from the World Bank (1985) and the Central Bank of Argentina have also been used for ir, r, iJ, AB, ODp, D0D9 OD, 0LP, OLG# defg, Ie, and IPrL/. All variables have been deflated by the wholesale price index (1974 - 100). The data on private wealth stock (W,,.) have been constructed by cumulatively adding the time series on private savings. Since linear functions are used, the error in estimating the initial wealth stock can be readily absorbed into the intercept term. The net profit of the banking sector ( nw) has been calculated as the residual of the financial system budget constraint2O/. While we attempted to 19/The variable r is defined as s rt r ((I + It)I(l+I't)) - 1 where it is the nominal interest rate. 20/ Onw (0LP/P) - Am + (bR/P) + (OLOIP) measure the expected rate of inflation (p0) using the adaptative and perfect expectations hypotheses21/. only the results obtained with perfect expectations are preseented in this paper. Perfect foresight expectations seeiA more approprlate than adaptative because the cost of Ignoring the future effects of current pollcy actlone can be qulte hlgh for economies that have a hiLstory of rapid lnflatLon. Assuming that economic agents refer to domestic condltions rather than to the foreign real interest rate, we defined the expected rate of return on foreign assets only wlth the rate of inflatlon, Data limltation enforced us to approximte the fimascash flow with the difference between potential and effective levels of productlon as measured by the Wharton index (IINDW). We assume that the enterprises could respond more easily to changes in desired investment when demand conditions are buoyant. Finally, in order to improve the specification of the model, we introduced into equations (4) an indicator of uncertainty about inflation (VE) measured as the variation of the rate of inflation22/ and we incorporated public investment (l3) into the private investment function (13) ao suggested by various authors231. In table 2 we report three-stage Least-Squares (3SLS) estimates for Argentina over the 1961-82 period. On the whole, the results are quite satisfactory. The explanatory power (I') and DV are both acceptable, suggestlng a good specificatlon of the model. The most interestlng aspect of these results concern the real rates of interest. Indeed, the estimated coefficients for r appear to be positive on financial savings (dAM/dr - 3.672) and negatlve on real savings (dOH/dr -1.234) and capital flight (ddJ/dr - -2.438). These results seem to be quite compatible with the Argentine experience, e.g. World Bank (1985). For instance, the deregulation of interest rates in 1977 resulted in a further increase in real interest for deposits and thus there was a dramatic increase of savings through the banking system. Our estimates suggest that the buildup of deposits was backed by repatriation of capital invested abroad and by decreasing demand for public bonds and capital goods. In short, an increase in the real rates of interest involved a portfolio shift from capital goods and foreign assets into monetary domestic assets. However, these estimates include only the direct effects of a rise in interest rates and, therefore, they cannot be interpreted as the total effect of such increase (see below for further explanations). Also of relevance is whether the banking system has played its intermediary role or not. The estimated impact of a variation in flow supply of credit (Ld) on private investment (dij/d(OLp/P) - 0.258) 21/This reflects the fact that, although rational forecasts could differ from actual price movements in stochastic models, rational expectations are equivalent to perfect foresight in deterministic models. 22/see Blejer (1979) and Gupta (1984). 23/For instance, see Blejer and Khan (1984) and Sundararajan and Thakur (1980). 16 Indicates that only a pare of banking credit h2s been used to finance productive Investment in Argentina. But, this is probably an underestimate of the full effect of financial liberalization on private investment because this policy can also increase the efficiency of investmont. In particular, higher intereot rates can improve the allocation of credit, thereby increasing the productivity of investmnt projects. AO an illustration, the relationship between the incremental output/capital ratio and the real interest rate on deposites appears positive and significant In the case of Argentina241. This suggests the existeace of a positive effect of an increase In interest rates on the quality of investment25/. The results pertaining to the other variables also deserve a brief explanation. The estimated coefflcient of Y is positive on financial savings and not asignificant on real savings. The effect of ADp/P seems to be positive on financial savings and negative on real savings. As equations are linearly dependane, we able to deduct that an increase in external debt led to capital flight during the 1961-82 period in Argentina (ddJ/(dODp/P)- 0.445). This positive correlation could be explained from the liquidity effect, corruption and policies which simultaneously promoted foreign borrowing and capital flight, e.g overvaluation of the local currency. The positive relationships between the flow supply of credit (OLP/P) and the demands for financial savings and for real savings are compatible with a priori theoretical expectations since the Argentinean private sector han been liquidity constrained. The expected rate of inflation (OQ) seerAs to have exerted a positive effect on financLal and real savings. With respect to real savings, this result corresponds to the one predicted by the theoretical analysis and, with respect to financial savings, it could be explained by the fact that banking deposits (in particular short term deposits) have been indexed in perLods of high and variable inflation. The negative effects of VE on domestic assets demand and money demaand confirm that uncertainty about inflation has increased capital flight. Finally, the estimation of the real private investment function seems to be satisfactory since all parameters have the expected signs, except the variables kp-, y1 and (ADp/P) whose effects do not appear significant. Note that public investment has a positive effect on private investment, suggesting that complementary relationships between both investment categories dominate in Argentina. Dynamic simulations (i.e. lagged variables are those generated by the model Ltself) show that the goodness-of-fit of the model as a whole is satisfying for Argentina over the 1961-82 period. Table 3 provides 24/We ussd the procedure proposed by Pry (1988) and Gelb (1989). The estimated coefficient for Argentina is equal to 1.206 on the 1961-82 perlod. 25/However, beyond the interest rates level, important factors such as the high degree of interlocking ownership and control that existed, and that continues to exist, between industrial firms and financial institutions may jeopardize financial liberalization policies in Argentina. 17 the correlation coefficient between historic and simulated serles and Theilos Inequality coefficient for the endogenous variables. If these coefficients indicate that the small structural model is stable and are indicative of its robustness, needless to say, that our simulation results have to be viewed with some care. Actually, they are intended primry as an illustration of our theoretical argumentso and not as perfect representation of the Argentinean economy. The model could be used as basis for deriving the short-run total response of private investment to an increase in real rates of interest. Since the model is linear, the siimulations results are independent of the starting conditions and, moreover, they are qualitatively independent of the absolute size of the shock. In Table 4 we report the total short-run elasticity of the main endogenous variables of the model to a one percentage point increase in the real rate of interest. Our simulation results indicate that real private investment is little responsive to int6rest changes (di /dr - -0.047), but this is not surprising since it reflects the presence of many opposing forces allowed for by the model. The total interest insensitivity of private investment is largely due to the portfolio shift from capital goods into financial savings. On one hand, the increase in monetary liabilities is favorable to private investment via the domestic credit market as suggested by McKinnon and Shaw. On the other hand, this portfolio shift undermines the private sector's willingness to hold capital goods and government bonds (d b/dr - -0.459). This last effect implies that the financing of the public deficit in the domestic credit market become less feasible and credits from the banking sysetem must be used to finace part of the deficit (d L./dr - 0.376). In order to offset the expansionary effect of large domestic credits to the public sector on money supply, the monetary authorities have to keep credits to the private sector under strict control. As a consequence, the resulting changes in private real investment depend critically on whether the Increase in domestic credits due to the McKinnon-Shaw effect exceeds or falls short of the reduction in dossetic credits due to the crowding out effect. Emplrical results seem to indicate that for Argentina the negative effect is higher than the positive effect (d L0Id- -0.284) so that the total effect of financial liberalization is weak (even negative) on private investment. 5. Conclud4n remarks The purpose of this paper has been to demonstrate with a simple structural model that a number of factors can influence the relationship between real Interest rates and private investment in LDCs. We emphasized on two specific problems. Pitat, we pointed out that the positive effect of a rise in domestic credit as suggested by McKinnon and Shaw could be offset by a portfolio shift from capital goods Into monetaMr assets. Second, we demonstrated that a financial liberalization policy could increase the demand of the public sector for credit extended by the domestic banking syetem, therefore limiting the funds avallable to the private sector, It Is noteworthy that this crowding out effect does not result from a change In the government's behavior, but rather from a shift in the portfolio of private agents. 18 Higher demand for bank deposits reduces the private sectorue willingness to hold government bonds so that the public sector is required to finance a given budget deficit with a larger amount of domestic credit. While the model has been estimated for Argentina, it is quite apparent that this specification can be readily applied to other developing countries as well. The general conclusion that emerges is that the effect of interest rates policies on the demand for capital goods Is weak in the case of Argentina, albeit the total impact might be stronger on the quality of investment than on the quantity16/. The absence of any strong relationship between real interest rates and the quantity of lnvesment does not result from exceedingly small direct interest elasticities of private investment. Instead, it is due to the Interactions of a number of mechanisms allowed for by the model which tend to neutralize the impact of such policies. Although the model used in this paper could be improved in mny ways, e.g. inflation should be endogenous and it is very unlikely that real interest rate policies can be considered as exogenous even if it rates are controlled rather than market-determined (e.g. Gelb (1989)), the policy implications of the exercise are straightforward. The increase In real Interest rates, which is a typical element of financial reforms, do not necessarily involve a positive effect on private investment unless the authorities are careful to ensure that s (1) Bank deposits are closer substitutes to unproductive assets (cash, gold) and foreign assets rather than to capital goods; (2) The financial sector assures an efficient allocation of domestic creditss (3) The flow of domestic credit to the private sector is not absorbed by the need of the public sector. BIBLIOGEAPHYS H. BLEJER, 'The Demand for Honey and the Variability of the Rate of Inflation s some Empirical Results', International Economic Review, 20, june 1979 H. BLEJER & T. TANZI, "Inflation, Interest Rate Policy and Currency Substitutions in Developing Economies s a Discussion of some Hajors Issues", World Development. vol.10, n.9, sept. 1982 H. BLEJER & H. KWAN, "Government Policy and Private Investment in Developing Countries', IMP Staff Payers, vol.31, n.2, June 1984. C. BRAINARD & J. TOBIN, 'Pitfalls in Financial Model Buildingo, American Economic Review, 1968 E. CONESA, "The Causes of Capital Flight from Latin America', Washington D.Cs Inter-American Development Bank, working paper, 1986. R CUMBY & R. LEVICH, 'On the Definition and Magnitude of Recent Capital Flight', in Capital Flight and_Third World Debt, D. Lessard and J. Williamson (eds), Institute for International Economics, Washington D.C., 1987 26/It could be clearly of interest In some future work to extend the analysis in this direction. 9 ~~~~~~~~~~~~19 J.T CUDDINGTON, "Macroeconomic Determinants of Capital Flight : An Econoetric Investigationa, In Capital Flihe and Third World Debt. D. Lessard and J. Williamson (edo), Institute for International Economics, Washington DC.o . 1987 R. DORNEBUSC & J. de PABLO, " Debt and Macroeconomic InstabliltY In Argetinag, The University Press of Chicago9 1988. S. FAZZARI, R. HUBBARD & B. PERTERSEN, Fi nancing Constraints and Corporate Investment", Brooklngs Papers on Economic ActLvity0 1, 1988 S. FISCHER, nAntlclpatione and the Nonneutrallty of Money", Journal of Political Economy. vol.87,n.2, aprll 1979. M. FRY, 1SavLngs9, Investment, Growth and the Cost of Financial Repression", World Development, vol.89 1980. M. FRY, "Money, terest, and Banklng in Economic Development6, The John Hopkins Unlversity Press, 19880 A. GELB, "Financial Policies, Growth, and Efficiency6, World Bank working Papers, 202, June 1989. A. GIOVANNINI, "The Interest Elasticlty of Savings in Developing CountrLees the Existing Evidence6, World Deve e, vol.11,n.7, 1983 K.L. GUPTA, "Finance and Economic Growth in Deyelopins Countries, Croom Helm, London 1984. N. U. HAQUE, "Flscal Policy and Private Sector Saving Behavior in Developlng Countries", IM Staff Papers, vol.35, n.2, june 1988 D. KHATKHATE, "Assessing the Impact of Interest Rates in Less Developing Countries6, World Development, vol.16,n.5,1988 R. McKINNON, 'Money and Capital in Economic Development6, Washington D.C., Brookings Institution, 1973. P. HONTIEL, "Credlt and Fiscal Pollcies in a Global Monetarist Model of the Balance of PaymentsO, IM Staff e, vol.31,n.4, october 1984 J. MORISSET,OL'affectation du financement externe dans un pays en vole de developpements un modele de portefeuille", Revue d'Economie Politioue, juillet-aout 1989. D. NICHOLS,"Some Principles of Inflatlonary Finance", Journal of Political Economy, n.2, march-april 1974. J.J. POLAK, "Financial Policies and Developmentu, Center for Development, OECD, paris, 1989. D. PURVIS, "Dynamic Models of Portfolio Behavior : more Pitfalls in Financial Model Buildings", American Economlc Review, LXVIII(3), 1978. 'E. SHAW, "Financial Deepening in Economic Development", New York, Oxford University Press, 1973. M. SIDRAUSKI, "Rational Choice and Patterns of Growth in a Monetary Economy", American Economic Revlew, vol.57, n.2, may 1967. S. SUNDARARAJAN, "Debt-Equity Ratio of FirLms and the Effectiveness of Interest Rate Policy t Analysis wlth a Dynamic Model of Savlngs, Investment and Growth in Korea", IM Staff Papers, vol. 34, n.2, June 1987. S. SUNDARARAAN & S. THAKUR, "Public Invesetment, Crowding Out, and Growths A Dynamic Model Applied to India and Korea", IM} Staff Papers, vol.27, n.4, december 1980. J. TOBIN, "Money and Economic Growth", Econometrica, 33, n.4, october 1965. U. TUN WAI & C. WONG, "Determinants of Prlvate Investment ln Developing Countrles", Journal of Development Studies, vol.19, october 1982. J. TYBOUT, "A Firm-Level Chronicle of Financial Crisls in the Southern Cone", Journal of Development Economics, 24, november 1986. 20 8. VAN VIJNBERGEN,"Credit PolLcy, Inflation ant Growth In a Financlally Repressed Economy0, Journal of DeveloMent Economics, Vol.13, august 1983. 0. VON 8 TENBERG, "The Uncertain Effects of Inflationary Finance on Growth In Developing Countries", Public Finance, n.2. 1983. THE WORLD BANK,, Arentinas Economic Memora8ndum", Washington D.C.o 1985. 21 Table 1s Expected Signs of the Parameters y va r (r*+Z) QDp/P OLP/P Ah 21 >0 eg £>0 R<0 e4<°0 0%>0 g20" e80 68>° a2*Ao eaS,C a,94'10- (KW% ,,2° >0 Table 2 s Estimates of the Model for Argentina (1961 -82) (t-statistics are In parenthesis) (4a) Om -0.652 + 0.069y + 1.0121 + 3.67r + 0.6730D,/P + (-0.36) (1.79) (1.96) (1.94) (3.47) 0.953bI/P - 0.012wp-1 - 1.238VE 12 0 0.954 (5.33) (-1.00) (-4.61) DV - 1.95 (4b),dh 5.063 + 0.063y + 1.447F - 1.23r - 0.118DP/IP + (4.07) (1.75) (4.67) (2.02) (-0.94) 0.678AL0/P - 0.023wp-1 0.418VE 12 0.935 (5.98) (-3.11) (-2.63) DV o 1.65 (13) ip C -4.959 - 0.OlSy-l + 9.229INDW + 0.258ALo/P P (-1.92) (-0.54) (3.58) (4.85) 0.201oD0/P + 0.091kp, + 0.448io R2 - 0.920 (-2.53) (0.79) (3.89) DV - 2.08 (6) (ALD/P) - m- (OR/P) - (OLO/P) + Onw (15) OL0IP - def0 - Ob O - /P (16) Ob O Oh - i, 22 Table 3 Comparison between Historic and Simulated Series (1961-82) Variables correlation coefficient Theil's coefficient lJlam 0.901 334 oh 0.825 0.314 6g o0.840 0.466 ip 0.780 0.203 OLp/P 0.757 0.544 ALgIP 0.687 0.497 Ob 0.770 0.532 table 4s Direct and Total ElastLcLtLes of the Real Rates of Interest, Effect Om Oh o i L OL,/P OL6/P Ob Dlrect 0.104 -0.034 -0.372 Total -0.073 -0.160 0.258 -0.047 -0.284 0.376 -0.459 1. Evaluated at sample means EBE Working Eaper Series Contact IiU AL4ho foLr f WPS690 The Terms-of-Trade Effects from the Gabor Oblath May 1991 J. Smith Elimination of State Trading in Soviet- David Tarr 37350 Hungarian Trade WPS691 Can Debt-Reduction Policies Restore Jacques Morisset May 1991 S. King-Watson Investment and Economic Growth 31047 in Highly Indebted Countries? A Macroeconomic Framework Applied to Argentina WPS692 Health Financing in the Poor J. Brunet-Jailly May 1991 0. Nadora Countries: Cost Recovery or Cost 31091 Reduction? WPS693 Report on Adjustment Lending II: Vittorio Corbo May 1991 A. Oropesa Lessons for Eastern Europe 39075 WPS694 Labor Markets in an Era of Susan Horton May 1991 M. Schreier Adjustment: An Overview Ravi Kanbur 36432 Dipak Mazumdar WPS695 Long Term Prospects in Eastern Ishac Diwan June 1991 S. King-Watson Europe: The Role of External Finance Fernando Saldanha 33730 in an Era of Change WPS696 Macroeconomics of Public Sector Jorge Marshall June 1991 S. Jonnakuty Deficits: The Case of Chile Klaus Schmidt-Hebbel 39074 WPS697 Volatility Reversal from Interest Paul D. McNelis June 1991 S. Jonnakuty Rates to the Real Exchange Rate: Klaus Schmidt-Hebbel 39074 Financial Liberalization in Chile, 1975-82 WPS698 Tax Policy Options to Promote Andrew Feltenstein June 1991 A. Bhalla Private Capital Formation in Pakistan Anwar Shah 37699 WPS699 Regulation and Deregulation in Ralph Bradburd June 1991 E. Madrona Industrial Countries: Some Lessons David R. Ross 37496 for LDCs WPS700 Trade Liberalization and the Oleh Havrylyshyn June 1991 N. Castillo Transition to a Market Economy David Tarr 37961 WPS701 Education and Adjustment: A Review Andrew Noss June 1991 C. Cristobal of the Literature 33640 WPS702 Should Price Reform Proceed Sweder van Wijnbergen June 1991 M. Stroude Gradually or in a 'Big Bang?" 38831 PRE Wor Paoer Series Contact AuIiQha for p1 e WPS703 The Polit;al Economy of Fiscal Sebastian Edwards June 1991 A. Bhalla Policy and Inflation in Developing Guido Tabellini 37699 Countries: An Empirical Analysis WPS704 Costs and Finance of Higher Rosemary Bellew June 1991 C. Cristobal Education in Pakistan Joseph DeStefano 33640 WPS705 What Causes Differences in Abby Rubin Riddell June 1991 C. Cristobal Achievement in Zimbabwe's Levi Martin Nyagura 33640 Secondary Schools? WPS706 Successful Nutrition Programs in Eileen Kennedy June 1991 0. Nadora Africa: What Makes Them Work? 31091 WPS707 Population, Health, and Nutrition: Population, Health, June 1991 0. Na 'zra Fiscal 1990 Sector Review and Nutrition Division, 31091 Population and Human Resources Department WPS708 Nongovernmental Organizations and Jocelyn DeJong June 1991 0. Nadora Health Delivery in Sub-Saharan Africa 31091 WPS709 An Empirical Macroeconomic Model Luis Serven June 1991 S. Jonnakuty for Policy Design: The Case of Chile Andres Solimano 39074 WPS710 Urban Property Tax Reform: William Dillinger June 1991 V. David Guidelines and Recommendations 33734 WPS711 Financial Reform in Socialist Millard Long June 1991 M. Raggambi Economies in Transition Silvia B. Sagari 37657 WPS712 Foreign Direct Investment in Thomas L. Brewer June 1991 S. King-Watson Developing Countries: Patterns, 31047 Policies, and Prospects WPS713 The Determination of Wages in Simon Commander June 1991 0. Del Cid Socialist Economies: Some Karsten Staehr 39050 Microfoundations WPS714 Women in Forestry in India Ravinder Kaur July 1991 A. Sloan 35108 WPS715 Promoting Girl's and Women's Rosemary Bellew July 1991 C. Cristobal Education: Lessons from the Past Elizabeth M. King 33640 WPS716 Financing Training: Issues and Christopher Dougherty July 1991 C. Cristobal Options Jee-Peng Tan 33640 WPS717 Does Financial Liberalization Really Jacques Morisset July 1991 S. King-Watson Improve Private Investment in 31047 Developing Countries?