Money, Debt, and Economic Activity

vent output prices and output from adjusting instantaneously. Both the size of deficits and the method of financing affect output and prices. Some principal implications are derived. Several of these are also demonstrated, using a graph to show the interaction of asset markets, output markets, and the financing of the budget deficit. Some main implications of standard analysis are rejected. The basis for several "monetarist" conclusions is shown.

and credit or bank-earning assets. This section presents a condensed version that omits the labor market, the determination of real and money wages, the interaction between output and labor markets, and the role of price anticipations. There is no growth in capital and labor force. A symbol dictionary (Appendix) defines the principal variables.
We assume, throughout, that costs of acquiring information and adjustment are smaller for the assets we consider than for output. Consequently, asset markets are cleared by suitable adjustment of asset prices within the time units relevant for our analysis. Output prices do not adjust rapidly enough to maintain equilibrium on the output market. One reason is that the adjustment of output and prices involves a production process that is slower and more costly than the adjustment process on the markets for money and credit. A second reason, that we do not pursue here, is that differences in costs of acquiring information give rise to differences in the prices anticipated by buyers and sellers.' Third, producers delay the adjustment of output and the labor force by allowing inventories to change. Once changes in expenditure are regarded as systematic, not random events, the adjustment of output and prices accelerates.
Three equations describe the interaction of real expenditure, real output and prices on the output market. Equation (1) describes the adjustment of real output of the private sector, y, to a discrepancy between aggregate real (private) output and aggregate real expenditure, d + g. Total expenditure is the sum of private (d) and government (g) expenditures on the output market. Equation ( (1) dt dd(i-APPi PW n Wh, e) di, d2 < 0; d3, d4, d5, d6 > 0; (2) p -p(y,K,y*) P1, P3 > 0; p2 < 0.
The remaining variables in the three equations are defined as follows: i is an index of nominal (or market) rates of interest; A, the anticipated rate of inflation; P, the price of existing real capital; Wn and Wh, the values of nonhuman and human wealth; e, the anticipated return on real capital per unit of real capital; K, the stock of existing real capital; and y*, anticipated real output.
The (partial) elasticity of p with respect to y measures the response of p to short-run changes in current output, holding expected output, y*, constant. We postulate that the response of p to short-run changes is less than the long-run response (Alchian and Allen 1967; Lucas and Rapping 1969;Phelps 1968); E(p,y) < E(p,y*). The size of the short-run response rises with the rate of capacity utilization, y*/K, and is higher in periods of sustained expansion than in periods of prolonged contraction. The shortrun response of y* to y depends on the factors determining producer's anticipations. It is sufficient for present purposes to assume that E(y*,y) does not exceed unity.
Three of the variables just introduced-Wn, Wh, and e-require additional description. Equations (4), (5), and (6) define these variables and introduce the financing of the government budget into the analysis. The presence of y* in equation (6) makes the anticipated yield on real capital depend on the factors affecting producers' shortand long-run anticipations. Changes in anticipations and particularly sudden, sharp, or autonomous changes in producers' market anticipations were emphasized by Keynes (1936) and Wicksell (1935)  The additional variables introduced in these equations are Cl, the ratio of the banking system's net worth to the monetary base;2 B, the monetary base; S, the outstanding stock of government debt (at face value); and v, the price per dollar of S. Our formulation of Wh and n assumes the prevalence of some stable distribution of real income.
The descriptions of the markets for credit and money follow our previously published work (1966,1968,1972) and are presented here with little elaboration. The equilibrium condition for the credit market, equation (7a), equates the banks' desired portfolio (aB) to the stock of earning assets offered to banks (a) .3 This equation proximately determines the 2 Real capital, K, does not include capital invested in the monetary system. The capital invested in the monetary system and the value of the banks' monopoly contribute to the positive value of o. Discussions following Pesek and Saving (1967) have suggested different methods of calculating the net worth of the banking system but agree that the value of a banking monopoly is part of net wealth. We treat o as a constant. 3 We assume throughout that the markets for money and credit adjust more quickly than the output market. The reason is that, in general, costs of acquiring information and costs of adjusting are substantially smaller for assets than for new production. The adjustment of output and prices, the h and p functions of our analysis, require more analytic foundation in terms of costs of information and readjustment than we have provided here. equilibrium value of the nominal rate, i. Equations (7b) and (7c) express the asset multiplier, a, and the stock of assets offered to banks, G, as functions of the main arguments of our model: aB -( 7a) aa(i, it, P, W., Wh, e) a,.. a5 > 0; a6 < 0; (7b) a = G(i-Jt Pi p Wn Wh, e, S) 01, G2 < 0; 63, G6, 67 > 0. (7c) The only variable introduced but not yet defined is it in (7b), the interest rate on deposits. The 6 equation is obtained by appropriate aggregation of the specialized credit markets for mortgages, corporate bonds, etc. Our hypothesis separates aggregative and allocative effects of portfolio allocations and assigns so little aggregative significance to the allocation of credit between submarkets that we dismiss the allocative effect on macrovariables. However, the stock of government securities is not eliminated in the aggregation; S enters (7c) with partial derivative equal to unity.
The credit market allocates the stock of government securities between banks and nonbanks and permits an individual wealth owner to adjust the composition of his wealth by borrowing or repaying loans and by buying or selling securities. Since the credit market assumes the role of (proximately) determining market interest rates most often assigned to the demand and supply equations for money, we assign to the money equation the task of (proximately) determining P, the price of existing capital and, implicitly, the net rate of return on real capital. Equation (8a), the equilibrium condition for money, is satisfied when the desired nominal stock of money held by the public, L, just equals mB, the nominal stock supplied by banks. The latter is the product of a money multiplier, m, and the monetary base. Equations (8b) and (8c) explain the money multiplier and L in terms of the arguments previously introduced: (8c) Several of the arguments of the credit and money equations have different effects on the two markets. The sign of the wealth elasticity of a is ambiguous and much smaller than the (positive) wealth elasticity of the L function. Also, the elasticities of the a function with respect to P and e are opposite to the signs of the elasticities of the L function. Several of the signs of (8b) are opposite to the signs of (7b). Stability of the system requires that the interest elasticity of the creditmarket equations exceed the interest elasticity of the money equations. The Hotelling conditions imply that this condition holds, and recent empirical studies either support or are consistent with the condition (see Brunner and Meltzer 1968;Zwick 1971 The portion of a deficit, Gt, financed by issuing base money or the portion of a surplus used to retire outstanding base money is denoted ,t; v is the change, dB,, resulting from central bank purchases or sales that are independent of the government's budget deficit or surplus-

Interaction of Prices, Stocks, and Flows on the Money Credit and Output Markets
This section develops some major properties of the system with the aid of a two-panelled diagram. One panel describes the output market and shows the relations between prices and output given by equations (2) and (3). The other panel shows the determination of partial equilibrium values of P and i obtained from equations (7) and (8). To bring out some main properties of the system, we start from a position of disequilibrium on the output market and trace the adjustment of p and y implied by the dynamics of the output market. We show that the adjustment of the output market variables disturbs the equilibria of the asset markets and changes the values of P and i that clear these markets. The changes in P and i, in turn, shift the expenditure (d + g) curve inducing additional changes in expenditure, output, and the price level. Throughout this section, we neglect the effect on the budget deficit of changes in prices and output.
The size and direction of the changes in p, y, P, and i and the comparative responses to these changes determine whether the adjustment process converges to an equilibrium and the speed of convergence. The slopes and the determinants of the shifts of each curve are written as elasticities. The expression E(x,z) denotes the elasticity of x with respect to z.
We begin in the output market. The negatively sloped line in panel 1 is the aggregate expenditure function, d + g. The slope of this line is given by the elasticity E(p,yld + g) and is derived from the expenditure function ,E(p, yld + g) The slope depends on four factors: (1) the ratio of the government's expenditure on output to total expenditure, ygid + g; (2) the responses of wealth and the yield on real capital to changes in expected output; (3) the response of expected output to actual output, E(y*,y); and (4) E(d,p), the price elasticity of real private expenditure. Each of the first three items is positive; the fourth is negative, so the denominator is negative.
The sign of the numerator depends on E(y*,y). In the short run, producers do not adjust production schedules to every change in expenditure, so E(y*,y) is less than unity. The numerator is, therefore, positive, and E(p,yId + g) is negative. In the long run, E(y*,y) 1, E(p,y d + g) falls, making the d-+ g line in panel 1 flatter and possibly upward sloping. cedure omits loans and repayments between the government sector and the private sector. Such transactions change the cash-flow deficit relative to the national income deficit and induce changes in relative prices and wealth and adjustments on asset and output markets. These operations are not adequately represented here. Other government activities, for example the activities of regulatory bodies, are also neglected.
The position of the d + g line depends on real government expenditure, g, the stock of real capital, K, the monetary base, B, the stock of government securities, S, the anticipated rate of inflation, a, and of particular interest here, on the interest rate, i, and the asset price of real capital, P.
Changes in P and i shift d + g vertically. The size of each shift depends on the ratio of two elasticities: E(p, Pld + g) -> 0; E(p, ild + g) <0.
,E(d, p) E(d, p) The bar on -(d,P), -(d,i), or other elasticities indicates a total elasticity; -(d,P) is the sum of two components, a partial elasticity and an induced change in wealth:

If the d-function is homogeneous of zero degree in all nominal values, E(p,Pld + g) is positive and less than one. The elasticity with respect to i has similar form: vS z(d, i) E(d, i) + E (d, Wn)E(V, i) < 0. Wn
Equiproportionate changes in P and i almost certainly shift the d + g curve in the same direction as the change in P. The reason is that E(P,P)

> -E(p,i)
by an amount that depends on two factors. One, E(v,i), is less than unity as long as the maturity of the debt is finite. (The debt is not entirely perpetuities). The other is the size of PK/Wn relative to VS/Wn; PK is considerably larger than vS. The short-run slope of the aggregate supply function, s, depends on E(p,y) of the price-setting function, equation (3). The long-run slope depends on E(p,y) + E(p,y*) and is steeper than the short-run slope. The position of the aggregate supply curve depends on y* and K as shown in equation (3). Increases in y* reduce s (for given p), and increases in K raise s.
Starting from any set of initial conditions, equations (1)-(3) determine aggregate expenditure, aggregate output, and a price level. Every combination is not an equilibrium position for the output market. Expenditure may exceed or fall short of output at the prevailing price level. The curves in panel 1 show an initial position of disequilibrium. Output is yo. The suppliers' behavior associates a price level, po, with this output. At this price aggregate expenditure is Yi, and excess demand is Yi -Yo. The suppliers' behavior, described by equation (1), implies that output increases by dy/y proportional to y,yo. This adjustment is a movement to the right along the supply curve in panel 1. Output price rises to pi, and excess demand falls. The system does not converge to the intersection of s and (d + g)1. The increases in p and y disturb the asset-market equilibrium and change tax collections and the size and financing of the government budget. These changes, in turn, change P and i, thereby shifting the d + g curve and inducing additional changes in the budget and its financing.

E (MM, P)
5 Note that the equilibrium is a partial equilibrium only. The asset markets are in equilibrium relative to the values of P and i and the prevailing p and y. If the values of p and y are not full stock-flow equilibrium values, generally there is an excess demand or supply of money and credit relative to the equilibrium p, y combination that clears the output market. When expected prices are included in the analysis, the partial equilibrium position of the asset markets must be defined relative to the expected prices. The expected prices may diverge from the actual prices and from the prices implied by the rate of inflation, ix, anticipated on the asset markets. The broken lines in panel 2 of figure 1 show an upward shift of CM and MM that moves the intersection to the northeast; P and i increase, and P increases relative to i. The effect of these changes on the output market is shown by the broken lines in panel 1. Since P increases relative to i and E(p,P d + g) > -E(p,ild + g), as noted earlier, real expenditure increases. The increase is shown by the position of the (d + g)2 curve. At price level P2, real expenditure is y:;, output is Y2, and excess demand is Y3 -Y2. The changes in y and p induce additional changes in P and i. Adjustment continues.

The numerators and denominators of the two elasticities consist of interestrate and asset-price elasticities of the demand and supply equations for money and credit. Each total elasticity includes the induced change in
The diagram cannot establish that the relative changes in P and i implied by the shifts of CM and MM are the changes implied by our hypothesis. Since the relative size of changes in P and i is a principal determinant of the direction of change in d + g, additional analysis is desirable. To show that, for dyly > 0, the shifts of CM and MM are those shown in the diagram, we analyze the components of (dPIP) (CM) and (dP/P) (MM) in more detail. There are three principal components: (1) the response to a change in output, the output effect; (2) the monetary effect of deficit finance; and (3) the debt effect of deficit finance.

The Output Effect
The output effect is the response of P and i to dyly. The first terms of (dP/P) (CM) and ( The denominators were shown above to be responses on the money and credit markets to changes in P. The numerators are, similarly, the responses of excess supplies on the money and credit markets to changes in y. Each numerator combines the responses of the demand and supply for money or credit to the changes in Wh, e, p, and y* resulting from the adjustment of the output market: Interest rates and asset prices generally rise in periods of expansion and fall during contractions. Two conditions are needed to assure this result. One is 1E(MMP)I > E(CM,P). The Hotelling conditions and the comparative effect of wealth on money and credit imply that this condition is met.6 The second is E(CM,y) > E(MM,y). We know from the discussion just above that within the range of capacity utilization observed during mild cycles, the sign of E(MM,y) depends on both E(L,e) and ,E(pj), so that E(MM,y) is likely to be negative and smaller than E(CM,y). This is the case shown in figure 1. However, it is not the only possible outcome under our hypothesis. In the recovery from a major recession or in the late stages of an expansion, the relative size of the shifts in the CM and MM curves or the direction of change may differ from those shown.

E(CM, y) E(y*, y) { [E , Wh) -E(a, Wh
Following a major recession, output is at very low levels relative to capacity, and E(p,y) is very small. Under these conditions, the shift of the MM curve, (dPIP) (MM) may be larger than the shift of the CM curve, (dP/P) (CM), so that interest rates fall and asset prices rise during the early stages of a recovers. As output rises relative to the fixed capital stock, c(p,y) rises and E(MM,y) falls; the size of the shift of the MM curve declines and the size of the shift of the CM curve increases; P and i once again rise and fall with dyly, as in figure 1.
In the late stages of an expansion E(py) eventually dominates E(MM,y), changing the latter elasticity from negative to positive and turning (dP/ P) (MM) negative. Interest rates rise relative to the asset price, P; in the limit, interest rates rise and P remains constant.
The conditions required for rising interest rates and constant asset prices cannot persist. Before the limiting point is reached, the effects of P and i on expenditure cancel. Expenditure remains stationary, and the output effect is exhausted. For any given K, B, and S, the output effect and the output adjustment produce a convergent movement of aggregate expenditure, and of the CM and the MM curves, toward a consistent, maintainable stockflow equilibrium. Any increase in y* contributes to the convergence by shifting the supply curve upward and to the left.
The output effect implies that, generally, asset prices and interest rates rise in periods of expansion and fall in contractions. Any effect of anticipations of inflation or of deflation on market rates and asset prices adds to the output effect and increases the size of the changes. However, even if anticipations of price change form and decay as slowly as some empirical evidence suggests, our hypothesis implies that changes in output can produce the observed pattern of changes in interest rates and asset prices by changing E(CM,y) relative to E(MM,y).

The Monetary Effect of Deficit Finance
The second group of terms in equations (i0a) and (lOb) makes the size and direction of the changes in CM and MM depend on the budget deficit and the portions of the deficit financed by issuing base money and bonds. In this and the following subsection, we hold output and the deficit constant and consider two polar cases; [= 1, the deficit is financed by issuing base money; and -0 the deficit is financed by issuing bonds. Later, we relax these constraints, allowing the deficit to change as output and prices change and combining the effects of financing the deficit with the output effect.
Issuing base money to finance the deficit increases the stocks of money and credit. The CM curve in figure 1 shifts to

The Debt Effect of Deficit Finance
Financing the deficit by issuing debt to the public shifts the CM curve to the right and the MM curve to the right. Both shifts have the same effect on market interest rates. Interest rates rise. The effect of the two shifts on asset prices cannot be determined unambiguously from equations (i0a) and (lob). The simultaneous solution of the asset-market equations shows that P rises in response to an increase in S under rather general conditions.8 If the effect of debt finance on interest rates is large relative to the effect on asset prices, the debt effect decelerates the adjustment of the output market. For the (partial) effect of issuing debt to reduce d + g, a more stringent condition must be met.9

In the introduction we listed a number of propositions that distinguish our framework from the standard IS-LM paradigm. The basis for several of these propositions is now clearer. One reason that the interest elasticities of the expenditure and demand for money functions are neither necessary nor sufficient for determining the relative responses to fiscal and monetary policies (proposition 1) is that changes in the base and in the stock of debt, whether the result of open-market operations or deficit finance, change asset prices and shift the demand for money, the expenditure function, and the asset-market curves. These shifts, a result of the interaction between the markets for assets and output, also explain why the real balance effect is not necessary or sufficient for a positive response of output to a change in the base (proposition 2). The dominant change in wealth results from the change in asset prices relative to the price of new output (proposition 3). We have also shown that a constant deficit financed by issuing debt raises market interest rates under our hypothesis (proposition 4). And we have worked throughout with a system in which the stock of moneycurrency and demand deposits is an endogenous variable dependent on the monetary base, interest rates, asset prices, and other variables. Yet, there is no point at which any main conclusion of our analysis would be altered if the stock of money was a constant multiple of the base and independent of any feedback from the output or asset markets (proposition 7).101 8The effect of S on i and P is given by E(i,SIAM) , [E(CM,P)E(L,W.)(vS/W -E(MM,P) I/den > 0, E(P,SIAM) -[E(CM,i)E(L,Wn) (vS/W.) -E(MM,i)E(G-,S) ]/den, where den is the determinant of the asset market matrix and is shown in the denominator of E(P,BIAAM) in the text. The den is negative: E(P,SIAM) > 0 if E(CM,i)/ Ce(MM,i) < (Wv/vS). ')The condition is e(d,P) [dP(S)/P] < -E(d,i) [di(S)/i], where dP(S)/P and di(S)/i are the effects of debt finance on P and i. 1%) One reason that the issue about the endogeneity of money persists is that the
The interaction of the asset and output market discussed in this section produces a movement toward an output-market equilibrium. We have shown that this movement converges and that the speed of convergence depends on the method chosen to finance the deficit, the choice of At. The larger the value of A, the larger the increase in money and the greater the size of the feedback from the asset markets to expenditure and output. The smaller the value of A, the slower the speed of convergence on the output market.
The choice of [t also affects the equilibrium position of the asset market, both directly and by changing the speed of convergence of the output market. It is clear that, with a given budget deficit and all other conditions unchanged, a low value of [ implies a slow process of adjustment to the full stock-flow equilibrium-a long lag in the adjustment of output to the deficit (proposition 6).
However, we have not followed the adjustment of the asset market to a new equilibrium or shown that both the asset and output markets converge to a consistent and sustainable equilibrium. In the following section we continue our discussion of the stock-flow interaction and develop the role of the budget more fully.

Shortand Long-Run Equilibrium
A disequilibrium in the output market disturbs the equilibrium of the asset market and sets off a process that moves the output market to a new equilibrium position. In the previous section we developed the response elasticities of the output-and asset-market equations that determine the direction and the speed of adjustment. Throughout that discussion we held the budget deficit constant and ignored the effect on the size of the deficit of changes in output, the price level, and other variables induced by the adjustment of the asset and output markets. Generally, we treated discussion rarely separates three distinct meanings of "endogeneity." One is the meaning used in the text. This interpretation implies that the feedbacks to the stock of money, through the effect of interest rates, asset prices, and other variables on the money multiplier, do not change the qualitative implications obtained from the analysis. Both time series analysis (Brunner and Meltzer 1968) and spectral analysis (Turner 1972) suggest that the feedbacks via the multiplier are small. A second meaning of "endogeneity of money" is that the base depends on current income, interest rates, or asset prices. At times, the argument is made that the central bank cannot control the base. The most common form of this argument applies to an open economy and states that short-term capital movements prevent a country from controlling the base. This argument implies that in the short run, dB., = -dB1. Studies by Fratianni (1971) and Neumann (1971) for Italy and Germany provide evidence that in these open economies, the feedback via dB. does not entirely offset dB1 within a year. A third meaning of endogeneity is that a central bank can only control the base if it relinquishes short-term "control" of market interest rates. This meaning improperly mixes "control" and "endogeneity" and focuses attention on the motives of central bankers. In our analysis, whatever central bankers do and for whatever reason they do it, the choices they make are entirely described by g, v and dB2.

each of the asset markets separately and did not make use of the properties of a simultaneous solution.
We now extend the analysis to include adjustments on the asset and output markets resulting from induced changes in the deficit. The response elasticities and the constraints required by our hypothesis move the system toward short-and long-run equilibrium positions. In this section we develop the dynamic implications more fully and analyze the properties of the equilibrium position. To bring out some main differences in implications, we reduce the asset-and output-market equations to two relations in the familiar i,y plane.
Panel 1 of figure 2 shows the asset market (AM) and output market (OM) relations as positively and negatively sloped curves, respectively.

The AM relation is obtained by solving the credit and money market equations simultaneously for i and P, at the prevailing level of output. Corresponding to each position of the output market whether an equilibrium or disequilibrium position-there are values of i and P that clear the asset markets. The slope of the AM curve, expressed as an elasticity, is given by E(i,y AM), the elasticity of i with the respect to y. The position of the curve depends on K, B and S: is yIAM) E(CM, y)E(MM, P) -E(MM, y)E(CM, P) E(CM, i)E(MM, P) -E(MM, i)E(CM, P)
The slope of the AM curve, E(iy AM), is the "output effect" on interest rates discussed in the previous section. We concluded there that output, interest rates, and asset prices generally rise and fall together, so a positive slope of the AM curve is expected. The denominator is negative. Its components are the four elasticities of excess-supply on the credit and money markets. A negative sign for the numerator is assured by two inequalities discussed previously: (1), IE(MM,P) I > E(CM,P); and (2), E(CM,y) > cE(MM,y) 1. The Hotelling condition implies the first inequality. The second inequality is generally expected to hold. Even if it does not hold, the numerator remains negative unless

1E(MM, y) (MM, P) E(CM, y) E(CM, P)
The main implication of this condition is that the AM curve is negatively sloped only if the demand function for money is considerably more responsive to the expected yield on real capital than to the price of existing real capital.11 This seems unlikely, and a positive slope seems assured at all values of y. The AM curve, therefore, shows a "flatter," but never a "flat," segment at low levels of output.
The AM curve is the closest analogy we can find in our system to the familiar LM curve. A principal difference between the two is that the slope of AM does not determine the size of the response to fiscal policy. Any change in government expenditure, or in any other variable that disturbs the equilibrium (or disequilibrium) position of the output market, also shifts the position of the AM line. The adjustment to the disturbance depends on the slope, or elasticity, c(i,y AM), but is not determined solely by this slope. The financing of fiscal policy changes B and S, and therefore P, and shifts the position of the AM curve.
The OM curve in figure 2 is a locus of equilibrium positions for the output market and is obtained by restricting our previous analysis in two ways. (1) Every point or the OM line is a position of flow equilibrium, y d +g.
(2) We replace P in the expenditure function with the solution for P obtained (jointly with i) from the simultaneous solution of the two asset-market equations. The substitution makes the slope as well as the position of the OM curve depend on the solution for P:
The denominator of E(i,ylOM) is negative. The sign of the numerator depends on three terms inside the braces. The first is negative. The second depends on E(P,ylAM), the "output effect" on asset prices that we discussed in the previous section. We noted there that the output effect is generally positive. However, we also noted that E(P,y AM) declines as output expands because rising capacity utilization raises one component, c(p,y). The increase in E(p,y) also increases the absolute value of the first term inside the braces. As a result, c(i,yJOM) becomes increasingly negative as output expands. The third group of terms is positive but relatively small in the short run. As E(y*,y) rises, the third term partly offsets the increase in the negative value of the first two terms.
The first term dominates the numerator if the private expenditure (d) function is homogeneous of zero degree in all nominal values. The reason is that homogeneity of zero degree implies that E(dp) -[E(dP) + c(dW,) + c(d,Wh)] > 0. We conclude that the numerator is positive. The OM curve is more negatively sloped at high than at low levels of output. If rising output is accompanied by rising anticipations of inflation, it increases. The increase in nt increases the change in interest rates obtained with a given change in output.
The slope of the OM curve does not fully describe the response to a change in the base resulting from monetary policy operations or the financing of a budget deficit. Every monetary policy operation and financ-ing of fiscal policy alters the equilibrium value of P that clears the asset market. Changes in B or S also shift OM by changing nominal Wn. These real balance and real indebtedness effects are weighted by (1 + 0) (B/Wa) and vSIW", respectively, and are, therefore, much smaller than the relative price and wealth effects induced by changes in P.
Each Our discussion of the AM, OM, and Gt curves brings out the interdependence of the system. Every change in the output market changes y and p, shifts the position of the AM curve, and changes the size of the budget deficit or surplus. Every change in the (partial) equilibrium of the asset markets changes output and the financing of the deficit. Every change in the deficit or surplus changes real expenditure, tax collections and the amounts of base money, and debt issued or withdrawn. These changes affect both the output and asset markets.
To simplify analysis of the interactions, we assume that the economy is closed, dB2 0; [t and g are constant, and v -0. We relax some of these restrictions presently.
Suppose some random event reduces output. Let the point io,yo in figure 2 be the position reached after the event. The output market is in dis-equilibrium. Output, yo, is less than yf, the level of output that maintains stock-flow equilibrium. At io,y0, output is also below the short-run equilibrium position of the output market given by the solid line, OMo. Since yo is below OMo, d + g exceeds yo at prevailing prices and interest rates, and output rises. If B, S, and K remain unchanged, adjustment proceeds along the AMo line toward the intersection with OMo.
The asset market is in short-run equilibrium at prices io,Po, and output yo. The decline in output from yf to yo reduces asset prices and market interest rates below the long-run equilibrium position at the intersection of AM1 and OM1. At output yo, the budget deficit is Do and is found by following the solid lines to Gt Do on panel 2.
Adjustment of the output market raises y and p. The dotted line from ioyo to the point A in panel 1 shows the direction of change in y and i. Note that the adjustment does not proceed along the AM line to the nearest OM line. The reason is that the financing of the budget deficit, Do, increases B and S, changing P and i and shifting the AM line to the right (direction of higher y). The size of the changes in P and i depend on the elasticities discussed in the previous section and on the choice of -i. The OM line also shifts to the right (direction of higher y).
The point A is an equilibrium position for the asset market and lies on an AM line (not shown) but not on an OM line. At A, real expenditure exceeds output; prices and output continue rising. The budget deficit is now Da, and requires smaller addition to the base and the stock of securities. The adjustment continues.12 Where does the adjustment come to an end? The AM curve continues to shift as long as new issues of debt and base money must be absorbed. Any change in the position of AM changes P and i. The OM curve cannot remain fixed if P changes, and P cannot remain fixed if output, the price level, or the budget change. Equations (i0a) and (lOb) above show that dPIP 0 when dyly 0 and Gt. In every position of stock-flow equilibrium, the budget must be balanced.
The broken lines of figure 2 show a stock-flow equilibrium. The budget is balanced, so there are no issues of debt and money to shift the position of the AM curve. With the asset market in equilibrium, P and i are constant, and the position of the OM curve is no longer disturbed by changes in P; with the output market in equilibrium, the asset market is no longer disturbed by changes in y and p. Once there is a full stock-flow equilibrium, the interest rate and level of output remain constant. In figure 2, this occurs only at the intersection of the broken AM and OM lines and with a balanced budget.
Both the final position reached by the economy and the speed of adjust-ment depend on the way in which the deficit is financed (proposition 5). If the deficit is financed by issuing debt, [i 0, the budget line in panel 2 moves to the right. If [t -1, the deficit is financed by issuing base money, and the budget line moves to the left. The total change of y and p from y(p( to the long-run stock-flow equilibrium, yfpf, increases as [t declines from one to zero. Since real resources are fixed, the effect of a larger issue of debt is a larger increase in p.*:1 The speed of adjustment also depends on the choice of At. The elasticity of output with respect to the base is a multiple of the elasticity of output with respect to the stock of securities. Hence, the response of y and p per unit time-and the speed of adjustment to equilibrium-declines as I falls.
A more formal demonstration makes the conditions for equilibrium somewhat clearer. There are five equilibrium conditions in our system: v, where Ft and v are policy variables and v is now defined relative to the initial period. Using these two equations to replace B and S reduces the number of endogenous variables to five, y,p,i,P, and D*. The necessary condition for long-run stock-flow equilibrium is satisfied.
We may now relax the constraints on B2, Ai, v, and g. In an open economy 1:8 The analysis implies that inflation or deflation can occur without any change in B. The size of the price change induced by a change in S depends on the elasticity e(p,y) and would be modified if we allowed for the effect on prices and output of capital accumulation and anticipations. But the choice of V affects the price level even if these effects are admitted. Price-level changes of this kind have not been important. Our analysis of inflation, presented at the Universities-National Bureau Conference on Secular Inflation, analyzes the issue in more detail and explains why most inflations or deflations have resulted from changes in money.
14 Our system requires adjustment to deal with a maintained inflation in which prices rise at a steady rate. Anticipations of inflation have to be included in the equations for the asset and output markets. Tax rates have to be included in the tax function so that tax rates can be reduced. If tax rates are progressive and taxes are not reduced, the government budget equations generate a surplus that disturbs the stock-flow equilibrium. The reason is that, with g fixed, pg rises at the rate of inflation and tax collections rise at a greater rate. y and d must be redefined to include exports and imports. With fixed exchange rates, financial flows induced by the balance-of-payments position change B2. Changes in B2 affect the system in much the same way as changes in B1, by shifting the AM and OM lines, changing P,ip,y, and D*. If policy makers change At, the composition of nominal wealth and the method of financing the deficit change. The AM and OM lines in figure 2 shift with changes in At, just as they do with constant i[t and the changes in B and S we have considered. In addition, the choice of [ affects the position of the Gt curve by changing i and S. The equilibrium prices of assets and output and the equilibrium interest rate also depend on the choice of [t. An increase in the (average) value of [ raises the price level and reduces the market rate of interest in the short run. Open-market purchases or sales that are independent of deficit finance change v, thereby changing B and S in opposite directions and, again, shifting the AM and OM curves. Increases or decreases in g change expenditure and therefore the prices of assets and output and the size of the budget deficit. Further consideration of the responses to B), It, v, and g, however, requires analysis of monetary and fiscal policies, deficit finance, and operating decisions of the central bank, issues that take us beyond the scope of the present paper.

Conclusion
The framework developed in this paper differs from the standard IS-LM framework in several principal ways. There are two asset markets and three prices-the prices of real assets, financial assets, and current output. Wealth owners are permitted to choose between money, bonds, real capital, and current expenditure. The real value of the outstanding stock of government debt does not equal the (discounted) present value of future tax liabilities. Costs of adjustment and costs of acquiring information prevent the output market from adjusting immediately. Excess demand or supply on the output market drives prices and output up or down but does not instantaneously restore equilibrium.
Differences in the framework produce differences in implications. Some of the principal implications are stated in the introduction and discussed in our preliminary conclusion above. Others concern the transmission of fiscal and monetary policies, the role of the credit market in the determination of interest rates, and the determination of equilibrium and disequilibrium values for prices and output.
We discard the notion that fiscal policies work "directly" while monetary policies work "indirectly." Both types of policy change the relative prices of assets and output. Relative price changes set off a process of adjustment that continues until a new stock-flow equilibrium is reached.
Interest rates typically rise in periods of economic expansion and fall in contraction. The IS-LM analysis explains cyclical changes in market rates by introducing anticipations of inflation or deflation. Our analysis recognizes that changes in the anticipated rate of inflation can explain the observed movements of interest rates, but anticipations are not required for the explanation. Changes in the banks' demand for earning assets (credit) and the public's supply of earning assets to banks offer an alternativeand to us a more credible explanation of the changes observed during periods of mild inflation or constant prices that characterize much of the peacetime history of the United States. A main failure of the standard analysis is the failure to determine both prices and output when the economy is not at "full employment" output. Keynes resolved the problem by making prices and money wages "rigid downward" so that his system determines real output when expenditure is below full employment output and determines prices when expenditure is above full employment. A central point of his analysis is that the distribution of the effect of monetary and fiscal policy between prices and output depends on the rate of capacity utilization, but the point is made only by denying any effect of price changes on output. Recent attempts to resolve the problem introduce costs of search, adjustment, and acquisition of information in the labor market and the supply function for output. Our analysis builds on these important developments, adds an explicit analysis of the asset markets and the interaction of stocks and flows, and offers a consistent explanation of stocks and flows, relative and absolute prices.