POLICY RESEARCH WORKING PAPER 1299 Fiscal Policy in Classical and In this model of cbssical and Keynesian open economies, Keynesian Open Economies bothpermanentafid transitory disturbances cause Klaus Schmidt-Hebbel changes in iong-lun output Klaues Schmidt-Hebbel Ltuis Serven and capacity - and transitory and permanent shocks may have opposite effects on thie current account. In the Keynesian economy, money- financed fiscal expansion causes real exchange rate depreciation, and non- money-financed fiscal expansion, appreciation '1'l,. Woerldl Iank; Ikollcv ResLarch Dcpu-m"1ent Mlacrocconotmiics and Growth DiVis;OI)X Mlav 1 994 PLICY RESEARCH WORKING PAPER 1299 Summary findings Sciimidt-Hebbel and Serven analyze the impact of fiscal lJsing parameters for a representative open econorny, policy changes in open economies, using a rational- the model is simulated to compare the dvoamic effects o.f expectations framework that nests two prototype increases in public spe?nding financed by taxatlimn, debt, economies: a neoclassical full-employment benchmark and money. The results illustrate four points: economy, with intertemporally optimizing consumers * Both permanent and transitory disturbances cause and firms and instant clearing of asset, goods, and factor changes in loiig-run output and capacity. markets; and a Keynesian economy, with liquidity * Transitory and permanent shocks may have opposite constraints and wage cigidity, which results in transitory effects on the current account. deviations from full employment. * liquidity constraints and wage rigidities tend to The model is forward-looking in that .he economy's amplify the cyclical adjustment to fiscal policy changes. short-run equilibrium depends or, current and * The Keynesian economy's response to fiscal shocks anticipated future values of all exogenous variables, and depends critically on the way the budget is financed: displays hysteresis (that is, its long-run equilibrium is money-financed fiscal expansion causes real depreciation; path-dependent). non-money-financed fiscal expansion causes appreciation. T'his paper - a product of the Macroeconomics and Growth Division, Policy Researcii Department - is part of a larger effort ill the dcpartment to model macroeconomic adjustment in open economics. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Emily Khine, room NI 1-061. extension 37471 (47 pages). May 1994. I 'I'he Policy Research Wtorking Paper Series disseminates the findings of u 0. This specification has the useful property that adjustment costs vanish in steady-state equilibiium - i.e., when gross investment per unit of effective labor is just sufficient to maintain the capital/effective labor ratio. The evolution of the latter is described by: (13) k = inv - (g + 8)k Market value maximization for unconstrained firms, as well as current profit maximization for constrained firms, yields the standard marginal productivity conditions for variable inputs (labor and imported materials).'" (14) Id = al v-l y (15) m = (I - a 2l - ) (e pmr)-l y Investment demand is, as described above, a combination of the market-value maximizing investment rule of unconstrained firms and the profit-constrained investment of restricted firms: (16) inv = - (g6 j+(1jp ~ 0 [2 [piq pilc g + (g + 6) k]I + (1 - 0 ) [p2 + igA where 0, is the share of non-constrained firms and 02 is the marginal propensity of liquidity-constrained firms to invest out of operational profits; 0 5 0 2 5 1. Unconstrained investment (the content of the first large right-hand side parenthesis) is derived from maximization of the value of the firm. This component of aggregate investment demand is geared to Tobin's maU& q -- i.e., average q minus the present value of the public investment subsidy per unit of capital". This reflects the fact that optimal investment is determined by the addition to future dividends of the marginal unit of capital, which excludes the subsidy due to its lump-sum nature. By ' The derivation of them conditons. an wei u of dte unconsauined componeot of invesment in equaiion (16) below, foUows die standard maxunzauon of the value of the firm, subject to equa_o (I 1) - (13), not preented here for brvity. II 'Me genwal remuom dat cause maginal nd averag q to diverge are spoUed out in Hayashi (1982). - 11 - contrast, the average value of existing capital (i.e., the present value of the dividends associated with an installed unit of capitai) must include the subsidy. In turn, investment by constrained firms (the last term in the right-hand of (16)) rises one-for-one with tlhe investment subsidy. The present value of the public investment subsidy is implicitly defined by the dynamic equation: (17) pvig = (r-g) pvig - piig Aggregate operational profits -- which determine capital formation by liquidity-constrained firms -- are defined as: (18) op = y - v -e pmr mr and dividends are the sum of operational profits, net of investment expenditure, the investment subsidy and the proceeds of new issues of equity: (19) d = op - piinv - pi iac + piig + q(k + gk) After determining aggregate investmnent according to equation (16), the second-stage investment decision involves allocating investment expenditure between domestic goods and imports, according to a Cobb-Douglas aggregation which renders constant expenditure shares: (20) in = y pi inv (21) im = (1-y) e p'|imv where y is the share of national-goods investment in aggregate investment expenditure, satisfying O s 5 1. Therefore the aggregate investnent deflator is a Cobb-Douglas average of national-goods investment prices and imported investment-goods prices: (22) pi = (epim)(l1Y 2.4 Conaumers Consumer preferences also allow two-stage budgeting distinguishing between intertemporal aggregate consumption decisions and intratemporal consumption composition choices. Intertemporal preferences reflect unit intertemporal elasticity of substitution (i.e., logarithmic intertemporal utility); intratemporal preferences also display unit elasticity of substitution between domestic and imported goods. - 12 - Private sector non-human wealth includes four assets: base money, domestic public bonds, foreign assets, and equity claims on the domestic capital stock. (23) a = hb + bg + e fbp + q(k-fe) - pvihb where the present value of money holding costs pvihb has to be subtracted from financial wealth; it is implicitly defined by the dynamic equation: (24) pvihb = (r-g) pvihb - i hb Human wealth is the present value of future labor income, net of taxes, and inclusive of current external transfers."2 Under the assumption that indiviouals can freely borrow against their future labor income at the going real interest rate, the path of human wealth is characterized by: (25) hu - (r-g)hu + [td - vI - e ftrp3 Consumption of non-liquidity constrained consumers is derived from standard maximization of intertemporal utility over an infinite horizon, subject to the intertemporal budget constraint equivalent of the private sector flow constraint in equation (3) -- which is exactly consistent with wealth definitions in equations (23) - (25). Solving the maximization problem yields the standard result that private consumption of unconstrained households is equal to the subjective discount rate (net of effective labor growth) times total (human and non-human) wealth." Unconstrained consumers are of course Ricardian, as they internalize the government's intertemporal budget constraint b)y anticipating the entire stream of current and future tax payments. B%.cause liquidity-unconstrained consumers face the same discount rate as the government'4, they are indifferent between tax, debt, or money financing. Therefore government debt -- although included in equation (23) -- ultimately "is not wealth" (Barro. 1974). Total private consumption demand is an aggregate of consumption by unconstrained and constrained consumers. with the latter consuming their current net labor income:"5 '" For expositionsl convenience, all txes and transfers have been lumped together in the human capital flow equation. Since both accrue in lump-sum fashion, dui is of no consequence for the model's properties. "As before, the analytical derivazions are stadard and can be omitted. ''Me assumption of equal discount rates is crucul for Ricardian equivalence to hold. Higher pnvate sector discount rates, whether due to finite lifetim (reflected by a given probability of death, as in Blanchard, 1985) or to a rsk premum on consumer.' debt relative to the borrowing cost of the government (e.g., McKibbin and Sachs, 1989) would cause Ricarchan equivalence to break down. '5 For discussion and enmprical analyses of the irplications of liquidity conSatiint for consumer behavior - as wel as for Ricardian equivalence - see, for example. Hayahi (1985), Hubbard and Judd (1986), Bernheim (1987), Leiderman and Blejer (1988), and Seazr (1993). - 13 - (26) cp = x1 [ (X2 g 8I)a j 2 - A) PC j + PC-X1) where 0 s A, s 1 is the share of unconstrained consumers, and X2 is the subjective discount rate. Disposable income is defined by: (27) yd = v I + e ftrp - td After determining aggregate private consumption levels according to equation (25), the second- stage private consumption decision allocates it to domestic goods and imports, according to Cobb-Douglas intratemporal preferences: (28) cnp = npc cp (29) cmp = -T) cp Le pcmp where 0 5 71 S 1 is the share of national-goods in aggregate private consumption expenditure. Therefore the aggregate private consumption deflator is a Cobb-Douglas index of national-goods prices and imported consumption goods prices: (30) pc = (e p=p)(l -11) 2.5 Govemmn The public sector could either determine policy exogenously or derive it from optimization of some objective function; for realism and simplicity, we choose the first option. Thus public consumption and investment expenditures are exogenously given. To finance its activity, the public sector can choose among taxes, money, domestic debt or external borrowing (or any combination of them). The accumulation of per capita real balances can be characterized as: (31) hb = [ng - (P/P) - g]hb where it is worth noting that the rate of money growth nmg will be endogenous under money finance of the deficit and exogenous otherwise. - 14 - 2.6 Foreigners The demand by foreigners for the domestically produced good is given by a conventional export function, which embodies imperfect substitution between the national good and the foreign final good and a normal relation to foreign income: (32) x = p1 (e px)P2 yfP2 where pV, P2, p3 2 0- Finally, the path of foreigners' equity holdings remains to be described. At every instant, foreign investors use dfi units of foreign currency (in real per capita terms) to purchase domestic shares, whose price in terms of domestic output is q. Hence their per-capita holdings of equity evolve according to the equation: (33) fe = e i g fe q In turn, profit repatriation equals the total volume of dividends earned by foreign investors, which is given by the product of the share of foreign-held equity and total dividends: (34) prem = fe d k 3. STEADY STATE AND STABILITY 3.1 The Steady State The long run equilibrium of the model is characterized by constant asset stocks in real per capita terms, constant asset prices (i.e., Tobin's q and the real exchange rate), and constant real wages with full employment. Thus, the government's budget must be balanced, and the current account deficit must equal the exogenously given flow of foreign investment, which in turn is just sufficient to keep foreign equity holdings (in real per capita terms) unchanged. Since the per capita real money stock is constant, long run inflation equals the rate of expansion of per capita nominal balances nmg-g. In turn, with a constant real exchange rate, domestic and foreign real interest rates are equalized by uncovered interest parity, and nominal exchange depreciation is determined by the difference between domestic and (exogenously given) foreign inflation. Hence, across steady states changes in the rate of money growth are fully reflected in the inflation rate (and thus in the nominal interest rate) and in the rate of nominal depreciation. By combining the model's equations, tne steady-state equilibrium can be reduced to two independent equations in the real exchange rate, real wealth, and the real interest rate: a goods market - 15- equilibrium condition, and a zero private wealth accumulation condition (in real per capita terms).16 Together they imply a constant stock of per capita net foreign assets. Real wealth accumulation can cease only when per capita consumption equals the per capita return on wealth. But the latter is just (rf-g) times the wealth stock (because of the assumption of perfect asset substitutability), while in the steady-state the former equals (XA-g) times the wealth stock (from (25)-(27))." Hence, this implies the well-known result that the rate of time preference X, must equal the exogenously given foreign real interest rate: (35) 2 - Rf Since (35) provides no information on the steady state wealth stock, we would be left only with the goods market equilibrium condition to determine both wealth and the real exchange rate-an obviously impossible task. This means that their steady-state values (and hence also those of all variables that depend on them) depend not only on the long-run values of the exogenous variables, but also on the initial conditions and on the particular adjustment path followed by the economy -- and therefore on parameters governing the speed of adjustment such as the degree of real wage rigidity or the magnitude of adjustment costs associated with investment. In other words, the model exhibits hysteresis. As noted by Giavazzi and Wyplosz (1984), this follows from the assumption of forward-looking consumption behavior derived from intertemporal optimization by infinitely-lived households with a constant rate of time preference and facing perfect capital markets. Nevertheless, certain important features of the steady state can easily be determined." On the production and investment side, long run equilibrium is characterized by full employment and a constant capital stock in per capita terms. From (13), gross investment is just inv = (g + 6)k, and adjustment ccsts are identically zero (from (12)). In turn, from (7), (17), (18) and (19), Tobin's q in steady state is given by: (36) q Fkipl (g+6) (rf-g) where Fk is the marginal productivitv of capital. If no firms are liquidity constrained (that is, B, = I in (16)), then (16) further guarantees that marginal q equals the price of capital goods or, equivalently, that average q equals the price of capital plus the unit investment subsidy (i.e. q = pi + pvig/k). Thus, from the above equation the marginal product of capital equals its user cost: (37) Fk = pi (rf+6) 16 See Serven (1994) for an analysis of the steady-state and dynamic properties of the non-monetary neoclassical vermon of our model. This is sictly correct only in the absence of liquidity constrints (XA = I). However, when XA < 1 wealth accumuluion would stUll equal (r - X,) times the wealth stock of unconsuained consutters (a + ), hu). Giavazzi and Wyplo:Z (1985) provide a method to solve analytically certain linear models with hysteresis. They show tdua the long-sun equilibrium depends on irntal conditions and on the speed of adjustment of the system. Since our model is nonlinear, however, a compnmble solution technique is not available. - 16 - Notice, however, that pi is an increasing function of the real exchange rate because of the import content of capital goods. In turn, for a given capitai stock Fk is a decreasing function of the real exchange rate, due to the use of imported materials in production. Hence, (37) defines an inverse relationship between the steady-state capital stock and the real exchange rate: across steady states, a real depreciation must reduce the capital stock, and (from (11) and (14)) also output and the real wage. It also follows that the long-run values of these variables depend, like the real exchange rate, on initial conditions and on the adjustment path of the economy. I9 What if some firms are liquidity constrained (i.e., l3, < 1) ? The negative long-run relationship between the capital stock and the real exchange rate is unaltered; however, in the steady state q does not equal the subsidy-inclusive price of capital goods, nor does Fk equal the user cost of capital. Provided the marginal propensity to invest of constrained firms (B2 in (516)) is not too large,a' the marginal product of capital must exceed the user cost, and Tobin's q must exceed the price of capital goods plus the Investment subsidy. Formally: (38) Fk= pi ((rf+8) + fl where f > 3 is a term that depends positively on the adjustment cost coefficient t and the rate of depreciation of capital, and negatively on B,, 82 and the investment subsidy.21 Tobin's q under liquidity constraints becomes: (39) q = pi f(rf+6) + + The intuition behind these results is simple: with binding liquidity constraints, firms cannot invest as much as they would want and therefore cannot close the gap between the shadow value of one additional unit of capital and its cost. This implies that, for a given long-run real exchange rate, liquidity constraints will cause the economy to achieve a lower capital stock and output, and a lower real wage as well, than in the fully unconstrained case. Using (37) or (38), the steady-state goods market equilibrium condition (4) can be rewritten as: 19 See Serven (1994) for further elaboution Tis is in contimt to similar dynamic modes (e.g., Sachs (1983), Giavam et al. (1982)), where capial goods have no import contot and thus the ueady-stat marginal product of capital (" wvll as the capital stock and real output) depends only on the relatve -price of mateials in terms of domestc goods (e pmr in our notaion). Hem the import con- of capital goods creae a negative relationship between the real exchange rae and ., even for a given real cost of imported inputs. Gavin (1991) ud Serven (1991) have shown tha this has unportant consequences for the effet of macroeconomic policies on inve 20' Te exact condition is 82 < (g+6)/(rf+6) - ig/(a2y). ' The exact expresion for f is f = X___f___ ________2____f__g - (rf+6), where Z * - i the 0, + (1-,) 2g (rf-g 62+--) public invennegt/pros output ratio. - 17 - (40) y(e,-..) T1 (rf-g)(a+hu) + cg + y pi (g+8) k(e,...) + x(e,...) which defines an inverse relationship between the long-run real exchange rate and real wealth: an increase in the real exchange rate (a real depreciation) generates excess demand for domestic goods and requires a fall in private wealth and consumption to restore market equilibrium.' As noted before, the particular levels of real wealthi and the real exchange rate that will obtain in the long run depend on the initial conditions and on the dynamic path followed by the economy. Aside from real wealth, the other key element in the determination of the long-run real exchange rate is the distribution of demand between the public and private sectors. Since public consumption is assumed to have no import content, an increase in cng creates excess demand for domestic gpods and leads to a real appreciation. As argued before, this would cause the capital stock and output to rise as well. An important implication of the model's hysteresis property is that transitory disturbances have long-run effects. For the case of fiscal policy, this has been recently highlighted by Turnovsky and Sen (1991).? In our framework this also means that even transitory monetary disturbances can have permanent real effects: if some consumers are liquidity constrained (or myopic), a transitory incrcea in inflationary taxation matched by a reduction in direct taxes will raise disposable income and consumpion, leading to reduced wealth accumulation and eventually causing a fall in long-run wealth and a peff ient real depreciation.' 3.2 Dypamics. Stability and Model Sotuion The precise dynamics of the model depend on the way the public deficit is financed. Under tax or money finance, the model is driven by ten dynamic equations. Four of them describe the time paths of predetermined variables: the capital stock, private foreign assets, foreign holdings of equity, and the real wage. At each moment in time, these variables are given by current and past values of endogenous and exogenous variables. Further, the four predetermined variables have to satisfy well-defined initial conditions. Under debt (domestic or foreign) finance, a fifth dynamic equation describes the time path of dte relevant debt stock. The remaining six dynamic equations describe the time paths of 'jumping' variables: Tobin's q, the real exchange rate, real money balances, human wealth, the present value of the investment subsidy, 2 Tis is guaanted Ly our assumption of conrtat expediture shar of domesic goods and impots in privae connipton Dt inveetmmnt With more general specification. ailowing lower substtuability etweenV dometic and forogn goods, a positive association between real wealth and tde real exchange rate in steady state (i.e., a 'conactionary devaluation' of dhe type analyzed by Krugnan and Taylor (1978)) could not in principle be ruled out I Turnovsky and Sen (1991) use a non-monetary model with intentemporiy optimizing consumers to show diat ansitory fiscal disurbances have long-run effects. Their result deperAs critically on the endogeneity of labor supply in thie frahework, wbic makes long-rn employment endogenous. In our cue, the dependence of tde long-run capitl sock on the real exchange rate ensures that ransitory fiscal shocks have permanent effecs despite the conancy of full eMployment acoss weady stae. I Without liquidity consaints, a monetary accelemaion (an increae in nmg) holding constant public consupohn would just amount to a change in the composition of txation between the inflation tax and (proent or futvre) direct taxes (or trafers), witaut any effect on wealth, conwumption, or any other real variable. - 18 - and the present value of the cost of holding money. They are not predetermined by the past and can react freely to 'news' about current and future values of exogenous variables; their equilibrium values at any point in time depend on the entire future anticipated path of the forcing variables. For the complete dynamic system not tc explode, these jumping variables have to satisfy certain terminal (transversality) conditions. Solving the model basically amounts to finding initial values for the non-predetermined variables such that, rollowing a shock, the model will converge to a new stationary equilibrium. The necessary and sufficient conditions for the existence and uniqueness of such initial values in linear models of this type have been investigated in the literature and will not be discussed -here, However, this is not the case for large nonlinear models such as this one.' While a formal proof of stability cannot be provided, numerically the model was always found to converge to the new long-run equilibrium under reasonable parameter values. The requirement that the predetermined variables satisfy initial conditions, while the jumping variables must satisfy terminal conditions, poses a two-point boundary-value problem, for whose numerical solution severai different techniques exist. One leading example is the "multiple shooting" method proposed by Lipton et al. (1982), which solves the model over a fixed time horizon starting from arbitrary ,uesses for the initial and future values of the jumping variables. The second is the "extended path" algorithm of Fair and Taylor (1983), which first solves the model also over a fixed time horizon, but starting from arbitrary guesses for the expected values of the jumpers, which are updated until they become sufficiently close tc the actual values obtained from the model's equations, and then gradually extends the horizon until the solution path is unaffected by the addition of more time periods. For the simulations below, we combine both techniques. First, we solve the model over an arbitrarily chosen time horizon using multiple shooting. To prevent the solution from being distorted by the choice of too short a time horizon (which would force the model to reach the terminal conditions too early), we then extend the horizon and recompute the solution path until the resulting changes in the solution trajectory of the endogenous variables fall below a certain tolerance,' at which time the process stops. In practice, the length of the simulation horizon required for this procedure to converge is strongly affected by two parameters governing the speed of adjustment of the system: the elasticity of real wages to employment (i.e., the slope of the augmented Phillips curve), and the magnitude of adjustment costs associated with investment. Finally, the model is made discrete for the numerical simulations, so for any variable x, x = x,1 - x. 4. SIMULATION; RESULTS This section discusses the dynamic response to external shocks, by presenting simulation results for the model introduced above. In a companion paper (Schmidt-Hebbel and Serven, 1993) we have D See BlanchLrd and Kahn (1980) and Buita (1984). 26 In pinciple, we could lirnearze the dystem around a steady gtate to detummi analytically the conditions under whici tie tramtion malnx posseues the saddle-pont property. For a tenth-order gysem, however, diis would be an inbsetrbbe task r We used a very act convergence cntenon, requLring thd tde maximum relative change between solutions in any varable at any time penod not exceed one-dhouaansz of one percent. Mis typically required a horizon between nxty Nd eighty perods for convergence. - 9 explored the model's response to a favorable foreign transfer shock (an external grant) and a favorable terms-of-trade shock (a decline in the price of the intermediate import used in production, say oil). We simulate the dynamic adjustment to fiscal and manetary policy shocks. The first-round magnitude of both shocks is common, equivalent to a 4% gain of initial steady-state output. We start by introducing the values of model parameters and exogenous variables and presenting the values of the endogenous variables at the initial steady-state equilibrium. Then we discuss the simulation results. 4.1 Model Parameterization and Initial Steady-State Solution Within the general structure spelled out above, two economies will be considered: (i) a neoclassical (NC) benchmark, and (ii) a Keynesian benchmark combining liquidity constraints and unemployment. Table 2 summarizes the common and distinct parameter values for these three economies. Under the neoclassical benchmark, liquidity constraints on consumption and investment are ruled out (B, = X, = 1.0). For the Keynesian case, the latter coefficients are reduced to 0.5. For the full-employment cases, the elasticity of real wage changes with respect to current employment is set at a very high level (w - 1,000) and indexation to lagged consumer-price inflation is ruled out (e = 1.0). By contrast, wage-setting behavior in the Keynesian benchmark gives rise to unemployment, as a result of a low employment elasticity (w = 0.25) and an important role of lagged inflation (O = 0.5). The latter feature reflects nominal stickiness of wages. Numerical values for other coefficients in the structural equations were borrowed from empirical estimates (Serven and Solimano, 1991, Elbadawi and Schmidt-Hebbel, 1991) and preceding simulation models (McKibbin and Sachs, i989, Giavazzi and Wyplosz, 1984) for various countries, complemented by estimates deemed to be representative for open economies. Table 2 also reports these parameter values shared by the three economies. Base money demand exhibits unit income elasticity. The interest semi-elasticity is 0.5, implying a seignorage-maximizing inflation rate of 200%. The share of labor, capital and intermediate imports in production (gross of imported materials) is 0.6, 0.3, and 0.1, respectively. The quadratic adjustment coefficient of investment is 2.5 andk the rate of capital depreciation is 0.04. The import component of aggregate investment is relatively large (0.4), exceeding that of private consumption (0. 1). The rate of discount of consumers is set at 0.06, which, according to equation (35), is also equal to the foreign real interest rate. Export demand exhibits a unitary foreign-income elasticity. The price elasticity of foreign demand for exports is 1.5. Before discussing the values of exogenous variables, a remaining question on model clos,ure has to be addressed: one residual endogenous variable for each of the two independent budget constraints remains to be chosen. For the simulations discussed below, the adjusting variable for the public sector will be either taxes (td), or domestic debt (bg), or money growth (nmg); and for the private sector the residual budgetary variable is foreign asset holdings (fbp).' The numerical values for exogenous variables are also based on both representative country magnitudes (as ratios to output) and previous models. Table 3 summarizes the exogenous variables for the initial steady state, common to the three economies. While the simulations show the response to changes in one exogenous variable (public consumption), all other exogenous variables are maintained at the levels summarized in table 3. Because in the initial steady-state domestic output per efficiency n Acul odl imulaios aume dim the pnvate sector intratmporal budget conatrain (eqution (3)) is the redundaa budget conswuit by Walras' Law, hence it is excluded from the set of model equaons. (Obvioualy the iuAtewporal budget conwmt is still used in deiving optml private consumpion leveb). Hence fbp is dte endogenous variable assocad to the extrnl ector budget consaint (2). - 20 - labor force unit is 1.0, all exogenous variables can be interpreted as ratios to initial steady-state output. Both the public and private sector benefit from foreign transfers, at 0.015 each. Foreign income is normalized at 1.0. Foreign direct investment flows amount to 0.005. Public indebtedness in foreign and domestic capital markets is 0.30 and 0.20, respectively. The sum of public consumption and investment is 0. 19, with a relatively large share of consumption. All absolute foreign price indices are normalized at 1.0, with zero foreign inflation. The rate of growth of the labor force in efficiency units is equal to the sum of population growth (2 %) and the rate of Harrod-neutral technical progress (I %). The foreign real (and nominal) interest rate is 6 %. Finally, nominal base money grows at 5%. The initial steady-state values of endogenous variables for the NC economy (and of most endogenous variables for the K)9 are reported in Table 4. Initial (and final) rteady-state output growth is determined by the rate of growth if the labor force in efficiency units (3%). Hence output per efficiency labor is constant; parameter values were chosen so that its numerical value is 1.0. At the initial steady-state equilibrium, total private sector (or consumer) wealth is almost 20 times output, corresponding to the sum of non-human wealth (4.990) and human wealth (14.417). The four components of non-human wealth (other than the exogenous public debt) are domestic base money (0.15), the domestic-currency value of foreign assets (0.874), the net value of equity given by the product of q (0.6) and the difference between the total c4pital stock (3.0) and the equity owned by foreigners (0.115), and minus the present value of costs derived from holding base money (0.4). Steady-state inflation at 2 % is given by the difference between money growth and output growth. Seigniorage is defined as the product of base money holdings and its rate of growth. At initial and final steady-state equilibria, seigniorage is 0.75% of output - the amount required to finance an operational public sector deficit of the same magnitude. Note that only at steady-state positions seigniorage is equal to the sum of the inflation tax (0.3% of output) and the growth effect on money demand (0.45% of output). At non-stationary equilibria, accumulation of money holdings drives a wedge between seigniorage and the latter sum. Initial steady-state private consumption is 0.58, mostly comprised by national-goods consumer spending. Stationary gross domestic investment is 0.21, all of which goes to replace depreciated capital per efficiency labor force unit. 60% of investment f-ls on domestically-produced goods. Investment adjustment costs are zero at the steady state, because they are only incurred on net investment. Exports are 0.20, intermediate imports are 0. 10, and total hnports reach a level of 0.242. The corresponding trade deficit of 0.042 and profit remittances (0.01) are financed by foreign transfers (0.03), the net return (net of growth) on foreign-held assets, which yields 0.017, and direct foreign investment flows (0.005). The latter flow finances an initial current account deficit (net of accumulation of foreign assets to maintain constant asset/output ratios) of 0.005. The steady-state nominal interest rate of 8% equals the sum of long-run domestic inflation and the real interest rate. At a al exchange rate of 1.0, all relative goods prices are also equal to 1.0. The 29 As disussd in secton 3.1, Tobin's q ts higher and investnent is lower under binding liquidity consamints than in the neoclAsl ecoomy. In fact, ina seady-ate values n the K economy ae 1.479 for q and 0.208 for gron domec inveetmeeg. which can be compared to the iaonary values in die NC economy, repot in Table 4. The saioary capitl ock is slighdy lower in die K economy (2.973), but the total equity value (q tunes k) is larger. Higher eqtuty moe than offsets lower foreig assets held by the private sector (equal to 0.853 in the K economy). Hence steady-state tota come weald and consumn are slighdy larger in die K economy. The stationary tred deficit is lighly lower in the K economy, as the retuir on foreigi ast holdinp bas lighdy deteriorated due to the lower aock of pnvate foreign ast hodinp. All otder variable reman unchaged in the K econory as comqaed to the NC case. - 21 - price of equity in units of national goods (q) is 1.444. Having normalized employment at 1.0. and with a labor share in production of 0.6, the real product wage is also equal to 0.6. TABLE 2: PARAMETER VALUES EOR SIMULATIONS Sae money demand - 0.16,0 - ' L., - -0.5 Watg settiing nrle - 1,000 (NC) or 0.25 (K), e - 1.o (NC) or 0.5 (K) - 0.91,a - 0.6, - 0.3 Production function = - 2.5 Investment adjustment costs 6 - 0.04 Physical capital deprocazon rate B- 1.0 (NC) o. 0.5 (K), 12 = 0.5 Private investment demand y 0.6 Domes ontent of investment XI 1.0 (NC) or 0.5 (K), X2 = 0.06 Private u imption demand l- 0.9 Domestic .cent of consufmption p = 0.2, p2 = 1.5, p, = I Export demand TABLE 3: INITIAL VALUES OF EXOGENOUS VARlABLES Income. Transfer and Capital Flows All Forewtn Price Levels 1.0 Foreign tmnsfer to public sector (ftrg) 0.015 Raw Foreign truanfer to pnvate sector (ftrp) 0.015 Foreign uncome (yf) 1.0 Populamion growth (pg) 0.02 Foreign direct investment (dfi) 0.005 Harrod-neutral technical progress (tg) 0.01 Foreign real interest (rf) 0.06 Stocks Nominal base money growth (nmg) 0.05 Domestic debt of public sector (bg) 0.2 Foreign assets held by public sector (e fbg) -0.3 Goods Flows Public nanioial-goods consumption (crip) 0.15 Public investment subsidy (ig) 0.04 - 22 - TABLE 4: INITIAL STEADY-STATE VALUES OF ENDOGENOUS VARIABLES Income, Capital and Transter Flows Ermlovmenc (1) 1.0 (erational profits (op) 0.300 Ouput (y) 1.0 Dividends paid (d) 0.260 Taxes itd) 0.183 Ranes Pnvae disposable inwome (yd) 0.433 Profit Rr.uttances (prem) 0.01 Nominal intenest rate on public debt (i) 0.08 Real interest rmte on public debt (r) 0.06 S_ocks Infiabon rate 0.02 Total pnvate sector wealth (a+hu) 19.407 All Reative Goods Prices 1.0 Non-humaz wealth of pnvate sector (a) 4.990 Stock of domesuc equity held by foreigners (te) 0.115 Other Prices Domemuc base money (hb) 0.15 Human wealth of private sector (hu) 14.417 Real equity price (Tobin's q) 1.444 Physical capWal (k) 3.0 Real wage per effetve labor unut 0.6 Present value of government investment subsidy (pvig) 1.233 Real exchange rate (e) 1.0 Present value of cost of holding money (pvihb) 0.40 Foreign asets held by pnvate sector (e fbp) 0.874 Goods Flows Privae aggregate consuimpuon (cp) 0.582 Private uipored-goods consumption (cnp) 0.058 Private national-goods consumpton (cnp) 0.524 Gross domestic investment (inv) 0.210 Private natonal-goods investment (in) 0.126 Private imported-goods investment (in) 0.084 Investnent adjustrnent costs (iac) 0 Exports (x) 0.20 Intermediae imports (mr) 0.10 Total imports (mr) 0.242 Trade balance -0.042 Current account balane -0,005 - 23 - The simulations below explore three alternative tyTes oi shocks: a permanent unanticipated (P) disturbance (hitting the economy from period I to terminal period T), a transitory unanticipated (TU) shock (bitting during periods 1 -4), and a transitoiy anticipated (TA) shock (bitting during periods 2-5). The simulations examine the effects of a fiscal expansion reflected by a public consumption increase of 4% of initial output, Ehree altrnative forms of financing are considered: lump-sum taxation (a balanced-budget fiscal expansion,, debt-financing, and monetary-financing. The discussion of the simulation results focuses on the deviations from an initial cady-state equilibrium (represented by period 0), distinguishing between the impact effects (in period 1) and the transition toward the new steady-state equilibrium (from period 2 to terminal period T). The discussion is based on the figures depicting the dynamic paths of the main endogenous variables. For tne tax- financed fiscal expansion, each figure page is divided into an upper panel, which reports the dynamic trajectories under the three types of shocks (P, TU, and TA) for the Neoclassical case, and a lower panel depicting the dynamic trajectories of the sarne shocks for the same variable in the Keynesian case. Because Ricardian eauivalence holds in the NC economy, the response of the key variables is the same regardless of th:.- financing alternative chosen. Thus, for the debt-financed case, we show the dynamic trajectories of fojur variables for the K case only. For the money-financed case, we show the dynamic trajectories of three variables for the NC economy and nine variables for the K economy. The terminal pe.iod T varies between 70, 80, and 90 periods. 4.2 A Balanced-Budget Fiscal Expansion Under a balanced-budget expansion, taxes are raised to finance higher government consumption and any other net expenditure increase arising endogenously, such as higher interest payments on the domestic public debt. The dynamic paths of the main endogenous variables in response to the tax- financed fiscal expansion are shown in figures 1-10. Ccnsider first the neoclassical (NC) economy and the case of a permanent tax-financed fiscal expansion. Tax revenue increases by 4 percentage points (pp.) of output iii period I and thereafter (Fig. I). A permanent increase in government consumption and taxes reduces wealth and private consumption accordingly (Fig. 2). i z t. anrd FconrorT1ic Growth Ross Levine 38526 WPS1286 Econo- [' rF anistormation and the Luca Barbone April 1994 S. Harbi Fiscal C lsis A Crit:cal Look at the Domenico ir. Marchetti 37143 Centr ' i%::.rcpean Experience of the 1990s WPS1287 Unstable Inflat on and Seignorage Jacques Morisset April 1994 D. Jenkins Rev, .:es in Latan Arnerica: How Many 37890 limes Can the Government Fool People? WPS1288 The Public Finance of Infrastructure: Vinaya Swaroop April 1994 C. Jones Issues and (otions 37699 WPS1289 A Fiscal Needs Approach to Equali- Anwar Shah April 1994 C. Jones zation Transfers in a Decertralized Federation 37754 WPS1290 Oil Price Instab:iity, Hedging, and an Stujn Claessens April 1994 D. Gustafson Oil Stabillzatiori Fund' The Case of Panos Varangis 33714 Venezuela WPS1291 A Survey ot Viet Nam's Legal Natalie G. Lichtenstein April 1994 M. Rangarajan Fra-:tivor, in' IT:'r 'a-1 -i 81710 Policy Research Working Paper Series Contact Title Author Date for paper WPS1292 Services as a Major Source of Growthi William Easterly April 1994 C. Rollison in Russia and Other Former Soviet Martha de Melo 84768 States Gur Ofer WPS1293 Product Standards, Imperfect Glenn Harrison April 1994 N. Artis Competition, and Completion of ihe Thomas Ruthertord 38010 Market in the European Union David Tarr WPS1294 Regulations, Institutions, and Hadi Salehi Esfahani April 1994 B. Moore Economic Performance: The Political 35261 Economy of the Philippines' Telecommunications Sector WPS1295 Why Higher Fiscal Spending Persists Bruno Boccara April 1994 M. Pfeiffenberger When a Boom in Primary Commodities Ends 34963 WPS1296 Earnings-Related Mandatory Salvador Valdes-Prieto April 1994 H. Rizkalla Pensions: Concepts for Design 84766 WPS1 297 How Relative Prices Affect Fuel Charles C. Guo May 1994 C. Jones Use Patterns in Manufacturing: James R. Tybout 37S99 Plant-Level Evidence from Chile WPS1298 Capital Goods Imports, the Real Luis Serven May 1994 E. Khine Exchange Rate, and the Current Account 37471 WPS1299 Fiscal Policy in Classical and Klaus Schmidt-Hebbel May 1994 E. Khinp Keynesian Open Economnies Lu s Server 37471 WPS1300 Dynamic Response to External Klaus Schmidt-Hebbel May 1994 E. Khine Shocks in Classical and Keynesian Luis Serven 37471 Economies WPS1301 Estimating the Health Effects of Bart Ostro May 1994 C. Jones Air Pollutants: A Method with an 37699 Application to Jakarta WPS1302 Sustainability: Ethical Foundations Geir 8. Asheim May 1994 C. Jones and Economic Properties 37699