I. Introduction

With the passage of the Economic Management Act in 1979 Iceland introduced a comprehensive system of monetary indexation. Bank deposits and loans were linked to price developments, while the Central Bank indexed the corresponding portion of required reserves. This policy had important effects on revenues from money creation, a source of Government revenues on which the Icelandic authorities had heavily relied. With a substantial part of base money now being insulated, the Government’s ability to derive revenues by responding to the inflation-induced demand for base money diminished. The tax base on which the inflation tax could be levied shrank substantially, as currency in circulation remained the only part of base money that was an interest-free liability of the public sector, and which could cover real expenditure through the issue of nominal liabilities. However, as a result of the indexation of financial assets in the banking sector, real interest rates increased significantly. With relatively higher opportunity costs of holding cash, the demand for currency decreased, thus further diminishing the inflation tax base.

On the other hand, the greater willingness of the public to absorb broad money and the subsequent increase in financial savings in the domestic economy meant a higher amount of seignorage for the authorities. However, since the inflation tax effect exceeded the seignorage effect, total revenues from money creation declined.

This decrease in monetization revenues has important consequences for fiscal policy which will be studied in greater detail in this paper. This will be done in the framework of the public finance approach to inflation. While, in principle, monetary indexation can affect the economy in various ways, ^2/ the analysis will be carried out under restrictive assumptions. For example, the implication that monetary indexation might have for the Government’s costs of nonbank borrowing—which could well aggravate the fiscal deficit—will be ignored. Insofar as profits in the banking sector may increase as a result of higher yields on required reserves and this effect leads to higher income tax revenues, this offset will also be neglected. ^3/

These limitations should be kept in mind especially with regard to the policy simulations presented in the last part of this paper. These simulations are preceeded in section II by a description of the evolution of monetary indexation and of the present situation. In section III the measurement of revenues from money creation will be discussed. In section IV the analytical framework will be presented. Finally, section V provides empirical results on the effects of indexation on the demand for money.

II. The Evolution of Monetary Indexation in Iceland

Monetary indexation in its present comprehensive form was institutionalized by the Economic Management Act of 1979, although there had been some moves in this direction in earlier years. In fact, the origins date from the mid-1970s when a policy of generally raising the cost of credit and revenues from monetary assets was introduced. In 1976, special interest premium accounts—with interest premiums to be regularly considered in view of price developments—were initiated, and, since 1977, a price compensation factor based on the cost of living and building cost indices had been added to bank lending and deposit rates. However, since these arrangements only provided for partial ad hoc adjustments of interest rates to inflation which started to accelerate at the end of 1977, bank deposits continued to decline in relation to GDP while the demand for credit rose unabated.

In order to diminish the disincentives to save associated with negative real interest rates, the Government announced in April 1979 a system of general price indexation of savings and credit to be implemented before the end of 1980. While earlier legislation explicitly prohibited the general use of indexation, the Economic Management Act of 1979 provided for full and automatic adjustment of interest rates to inflation.

The first steps toward positive real interest rates were taken during the remainder of 1979 when nominal interest rates were raised three times. ^4/ Banks were given permission to grant fully indexed loans carrying a positive real rate of interest with the loan principal indexed to the credit terms index. ^5/ The minimum loan maturity was initially set at four years, shortened to 2 1/2 years in early 1981, and to six months in 1983. The real interest rate on indexed loans was initially set at 2 1/2 percent. On the deposit side, banks began to offer fully indexed two-year time deposits carrying a real interest rate of 1 percent in 1980, and in 1981, the minimum maturity was reduced to six months. Three-month indexed deposits with a zero real interest rate were also made available in 1982.

Real interest rates were initially set by the Central Bank (as were the nominal rates on nonindexed instruments), but since 1984 market forces have been given an increasing role in interest rate determination. To maintain competitiveness, the banks introduced “switching term deposits” in 1984, where the terms switched between ordinary and indexed rates according to which was more favorable to the depositor. With the new Central Bank Act of 1986, commercial banks were freed to set practically all their own interest rates. The deregulation of interest rates caused deposit and lending rates to rise, as banks increasingly competed among themselves and with other institutions for financial savings (Chart 1). The ratio of total deposits to GDP increased considerably in response to higher real interest rates, with switching terms deposits proving to be particularly popular (Chart 2).

Deposits in Deposit Money Banks

(In Percent)

Citation: IMF Working Papers 1990, 020; 10.5089/9781451921939.001.A001

Source: Information provided by the Icelandic authorities.1/ In percent of total deposits.2/ Switching-term deposits, introduced in 1984, are deposits where the terms switch between ordinary and indexed rates according to which is more favorable to the depositor.3/ End of June data.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Deposits in Deposit Money Banks

(In Percent)

Citation: IMF Working Papers 1990, 020; 10.5089/9781451921939.001.A001

Source: Information provided by the Icelandic authorities.1/ In percent of total deposits.2/ Switching-term deposits, introduced in 1984, are deposits where the terms switch between ordinary and indexed rates according to which is more favorable to the depositor.3/ End of June data.

• Download Figure
• Download figure as PowerPoint slide

Deposits in Deposit Money Banks

(In Percent)

Citation: IMF Working Papers 1990, 020; 10.5089/9781451921939.001.A001

Source: Information provided by the Icelandic authorities.1/ In percent of total deposits.2/ Switching-term deposits, introduced in 1984, are deposits where the terms switch between ordinary and indexed rates according to which is more favorable to the depositor.3/ End of June data.

• Download Figure
• Download figure as PowerPoint slide

Real Interest Rates ^1/

Citation: IMF Working Papers 1990, 020; 10.5089/9781451921939.001.A001

Source; Central Bonk of Iceland, Economic Statistics Quarterly.^1/ 1972-79 data points refer to annual series, and 1980-89 ore quarterly.2/ Deflated by the change in the credit terms index during past 12 months.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Real Interest Rates ^1/

Citation: IMF Working Papers 1990, 020; 10.5089/9781451921939.001.A001

Source; Central Bonk of Iceland, Economic Statistics Quarterly.^1/ 1972-79 data points refer to annual series, and 1980-89 ore quarterly.2/ Deflated by the change in the credit terms index during past 12 months.

• Download Figure
• Download figure as PowerPoint slide

Real Interest Rates ^1/

Citation: IMF Working Papers 1990, 020; 10.5089/9781451921939.001.A001

Source; Central Bonk of Iceland, Economic Statistics Quarterly.^1/ 1972-79 data points refer to annual series, and 1980-89 ore quarterly.2/ Deflated by the change in the credit terms index during past 12 months.

• Download Figure
• Download figure as PowerPoint slide

Overall, indexation has had predictable effects. Depositors responded to the more favorable terms by increasing their financial savings. ^6/ Index-linked borrowing has also gained in popularity. At end-1988, 56 percent of all bank deposits were linked to the credit terms index, 7 percent were foreign-exchange linked, ^7/ and 37 percent were non-indexed. On the credit side, 35 percent of bank loans were linked to the credit terms index and about 31 percent were foreign-exchange linked, leaving about 34 percent nonindexed. For the financial system as a whole, only about 15 percent of loans are nonindexed. Moreover, almost all treasury bonds are indexed, as are most private sector bonds.

However, recently the Government has attempted to reduce financial indexation. In July 1988, it decided to ban direct indexation of the principal value of financial instruments with maturities shorter than two years, and to set the minimum maturity for indexed bank deposits at six months. In June 1989, the Central Bank announced that indexation of switching term deposits would only apply to the unused amount in each savings account in each six-month period.

When indexed time deposits were introduced, similar accounts were opened in the Central Bank for deposit money banks’ required reserves. While originally only the portion of required reserves corresponding to the banks’ indexed deposits was indexed, ^8/ in February 1987 it was decided to index all required reserves at a zero real rate of interest. ^9/ However, since March 1988, required reserves have earned a real interest rate of 2 percent. ^10/

III. The Measurement of Revenues from Money Creation

The Icelandic financial system has undergone substantial changes in the last decade. In response to the threat that the highly regulated system of financial repression could destroy itself (Eggertsson, 1989), successive Governments have introduced far-reaching measures to reform the financial and money markets. While these markets were still embryonic at the end of the 1970s (Eggertsson, 1982, p. 1), there have indeed been remarkable developments in some areas. The most important reform ^11/—the introduction of monetary indexation—has generated predictable effects. Depositors have responded to more favorable terms, and domestic financial savings have increased.

If required reserves which have correspondingly increased had not been indexed, revenues from money creation would have become larger. In the steady-state, these revenues are equal to (g + π)m, where g denotes the growth rate of real income (assuming unitary income elasticity), π the inflation rate and m real base money. While gm can be interpreted as seignorage, that is, the steady-state increase in the real value of base money, πm represents the inflation tax, i.e., the amount of nominal balances that needs to be accumulated just to keep the real value of base money constant (Anand, van Wijnbergen, 1988, 1989). Assuming that the Central Bank pays no interest on its liabilities and issues monetary assets at a rate equal to the inflation rate, it taxes money holders (the DMBs which pass the loss on their depositors) at that rate (Fischer, 1982). ^12/ However, in the case of Iceland this assumption cannot be maintained. As noted above, all required reserves earned interest before 1979, while after the introduction of indexation that portion of required reserves corresponding to the DMBs indexed deposits was also indexed. ^13/ From February 1987 to February 1988, all required reserves were fully indexed but received zero real interest. Since March 1988, real interest of 2 percent has additionally been paid. Taking this into account, it appears more appropriate to use the following measure of revenues from money creation (RM): ^14/

Equation (1) follows an approach suggested by the OECD (1988, p. 65), where revenues from money creation in Iceland were calculated by the (negative of the) real interest rate paid by the Central Bank: (i - π)m . However, this approach does not take into account that interest is paid only on a fraction of base money, namely required reserves (1 - ϕ)m, whereas currency in circulation, ϕm, is indeed an interest-free liability of the Central Bank.

Calculations show that revenues from money creation have substantially decreased in the period of indexation. According to Table 1 the average annual revenues were 2.96 percent of GNP in the pre-indexation period from 1971 to 1980, whereas the period from 1981 to 1987 the average revenue was 1.49 percent.

Table 1.

Revenues from Money Creation

(In percent of GDP)

Source: Staff calculations.

Table 1.

Revenues from Money Creation

(In percent of GDP)

	Seignorage	Inflation tax revenues	Revenues from money creation
1971				1.45	0.64	2.09
1972	0.60	1.58	2.19
1973	0.57	2.50	3.07
1974	0.47	3.32	3.79
1975	0.03	2.80	2.83
1976	0.42	2.28	2.70
1977	0.78	1.63	2.41
1978	0.50	2.53	3.03
1979	0.47	1.87	2.34
1980	0.34	4.90	5.24
1981	0.22	2.16	2.38
1982	0.11	1.71	1.82
1983	-0.24	3.09	2.85
1984	0.23	0.90	1.13
1985	0.16	0.32	0.48
1986	0.47	0.72	1.19
1987	0.38	0.23	0.61

Source: Staff calculations.

View Table

Table 1.

Revenues from Money Creation

(In percent of GDP)

	Seignorage	Inflation tax revenues	Revenues from money creation
1971				1.45	0.64	2.09
1972	0.60	1.58	2.19
1973	0.57	2.50	3.07
1974	0.47	3.32	3.79
1975	0.03	2.80	2.83
1976	0.42	2.28	2.70
1977	0.78	1.63	2.41
1978	0.50	2.53	3.03
1979	0.47	1.87	2.34
1980	0.34	4.90	5.24
1981	0.22	2.16	2.38
1982	0.11	1.71	1.82
1983	-0.24	3.09	2.85
1984	0.23	0.90	1.13
1985	0.16	0.32	0.48
1986	0.47	0.72	1.19
1987	0.38	0.23	0.61

Source: Staff calculations.

There are several reasons for this, the most important being the move to non-negative real interest rates on required reserves. As discussed, the bulk of base money in Iceland is not an interest-free liability of the Central Bank. While in the 1970s interest was paid on required reserves at a rate equivalent to the interest rate paid by the DMBs on deposits plus a certain premium, the real interest rate on required reserves was still negative. To the extent that the inflation rate exceeded the nominal interest rate, the Central Bank received inflation tax revenues (in addition to revenues arising from the issue of bank notes and coins). When indexation was introduced, however, the portion of required reserves corresponding to the DMBs indexed deposits was also linked. Thus, the base on which the inflation tax could be levied shrank, irrespective of the increase in deposits in DMBs and hence in required reserves. When it was decided in February 1987 to index all required reserves, currency in circulation remained the only source of inflation tax revenues.

When indexed deposits were introduced, the opportunity cost of holding cash rose. While financial savings in bank deposits increased, currency in circulation (in relation to GNP) declined from 2.17 percent in the period form 1972 to 1980 to 1.25 percent in the period 1981 to 1987. This substitution effect has also diminished the inflation tax base, leading to a decline in revenues from monetization. Two further effects should be mentioned. One is the reduction of the (average) reserve requirement ratio by 10 percentage points in April 1985 and by 5 percentage points in March 1987. ^15/ The other noteworthy effect was the deregulation of interest rates in the 1980s. To the extent that increasing interest rates on deposits in DMBs also led to higher interest rates on required reserves and hence to a smaller gap between nominal interest rates and the rate of inflation, the deregulation contributed to the decline of revenues from money creation.

As a consequence of this decrease in revenues from monetization total Government revenues have ceteris paribus been smaller than they would have otherwise been. This requires a cut in the deficit or else public debt will rise (Dornbusch, 1988, p. 12).

IV. Revenues from Money Creation in the Framework of the Public Finance Approach to Inflation

In this section, the analytical framework is presented in which the effects from decreased revenues from money creation can be studied in greater detail. Here, we follow the public finance approach to inflation mainly developed by Phelps (1973), Dornbusch (1977), Buiter (1983), Drazen and Helpman (1987). Without denying that, in the short run, demand pressures or cost-push factors may be more important determinants of inflation, this approach argues that the ultimate reason for the existence of high inflation rates lies in the Government’s need for revenues. As Anand and van Wijnbergen (1988, 1989) have shown, the public finance approach provides a useful apparatus for deriving the implications of financial sector reforms for fiscal policies. While Amand and van Wijnbergen (1988) focused above all on changes in bank deposit rates, their model can easily be adjusted to take account of the effects of monetary indexation. In the following the analytical framework is presented which will later be applied to different questions.

In order to derive the fiscal consequences of monetary indexation we start from the following public sector deficit identity:

On the left-hand side of equation (2) are the expenses of the public sector (net of tax revenues or transfers from the Central Bank): ^16/ the noninterest deficit D, plus nominal interest payments on domestic and foreign debt. The variable i (i^*) is the nominal domestic (foreign) interest rate on domestic (foreign) debt B (B^*), E is the nominal exchange rate. On the right-hand side are the financing items, i.e., the issue of domestic or foreign debt plus Central Bank advances to the public sector.

Taking into account that the Government can shift a part of its deficit into the Central Bank accounts by simple changes of bookkeeping practices and that Central Bank credit to the Government represents a claim of one public entity on another, it seems useful to consolidate intergovernmental agencies’ debt. According to equations (3) and (4) derived from a simplified Central Bank balance sheet

base money is equal to currency in circulation (Cu) plus required reserves held by the DMBs at the Central Bank. It is issued to cover credit to the Government and the Central Bank’s accumulation of net foreign assets (NFA^*E), insofar as this is not already covered by the Central Bank’s accumulated profits of net worth (NW). ^17/ Subtracting the Central Bank’s profits (interest earnings on foreign reserves) from the budget deficit and the Central Bank profit’s counterpart, i.e., the increase in net worth, from the public sector’s increase in liabilities yields

By consolidating the change in the Central Bank’s net foreign assets and the change in the Government’s foreign debt we obtain

Taking into account that exchange rate changes imply capital losses or gains which are part of the cost of servicing foreign debt, and deflating all variables, equation (7) can be reformulated as follows:

where lower-case letters now denote real variables.

According to equation (7) the fiscal deficit, including the Central Bank’s profit- and loss-account, but only counting real interest payments, equals changes in the real value of domestic and foreign debt and base money. If no nominal interest is paid on required reserves,

Δm represents the Government’s revenues from money creation, which is equal to (g + π)m in the steady state. Since this assumption cannot be maintained in the case of Iceland, Δm has to be replaced by equation (1).

In order to evaluate different policy options it is important to know how the demand for money responds to changes in inflation and economic growth and whether these reactions are different, if monetary asets are indexed. In order to seperate effects on the demand for currency (C) and for deposits (D) the following partial adjustment money demand functions will be estimated: ^18/

where P is the GDP deflator of period (t), Y^r real GDP; i a weighted average of interest rates on demand, time, and savings deposits; IT the inflation rate; and e an error term.

Finally, it seems reasonable to assume that the alternative sources of financing a budget deficit are subject to certain limitations. Here, a particularly simple debt strategy is supposed, i.e., fixed debt-GDP ratios for both internal and external debt. ^19/ Target values for these ratios imply that real domestic debt cannot grow faster than real output ^20/ and real foreign debt cannot grow faster than the product of output times the real exchange rate:

Inserting (9a) and (9b) in (6) yields (expressed as a percentage of GDP):

According to equation (10) the noninterest deficit plus real interest payments on domestic and foreign debt cannot exceed what can be financed through debt issue at target debt-output ratios, plus the revenue from (steady-state) money creation.

V. The Demand for Money: Empirical Results

Using quarterly data ^21/ the partial adjustment model given in equations (8a) and (8b) was estimated for the pre-indexation period 1972 (2) to 1980 (2). The regressions produced reasonable results which are reported in Table 2.

Table 2.

Regression Results ^1/

Source: Staff calculations.

^1/t-statistics in parentheses; one asterik indicates that the coefficient is significantly different from zero on the 10 percent level, two asterisks indicate significance on the 5 percent level and three asterisks on the 1 percent Level.

^2/Lagrange Multiplier Test for Auto Regression Conditional Heteroscedasticity; critical values at the 5 percent Level in parentheses.

^3/Lagrange Multiplier Test for Residual Autocorrelation; critical values at the 5 percent Level in parentheses; according to further tests the hypothesis of the non-existence of residual autocorrelation of different orders could not be rejected.

^4/Chi ² - Test for normal distribution of residuals; critical values on the 5 percent level in parenthesis

Table 2.

Regression Results ^1/

Dependent Variable	Period	Independent Variables						R2	Durbin-Watson	ARCH(4) 2/	$\mathrm{AR}{\left(4\right)}_1^4$3/	NORM(2) 4/
		Constant	$\mathrm{L}\mathrm{og}{\mathrm{Y}}_{\mathrm{t}}^{\mathrm{r}}$	it	π	Log (Ct-1/Pt-1)	Log (Dt-1/Pt-1)
Log (Ct/Pt)	1972(2) -1980(2)	-3.481 (-1.370)*	0.4195 (1.543)*	-0.775 (-1.507)*	-0.204 (-1.609)*	0.666 (5.960)***	—	0.75	2.05	0.28 (0.884)	0.124 (0.328)	1.138 (5.991)
Log (DtPt)	1972(2) -1980(2)	-7.509 (-3.171)***	1.069 (3.774)***	1.051 (2.289)**	-0.259 (-3.280)***		0.341 (2.583)**	0.84	1.89	0.58 (0.684)	0.0318 (0.0343)	1.123 (5.991)
Log (Ct/Pt)	1980(4) -1987(4)	-9.091 (-3.080)***	1.043 (3.186)***	—	-0.086 (-1.362)*	0.412 (2.618)**	—	0.86	1.95	0.51 (0.729)	0.124 (0.328)	0.59 (5.991)
Log (Dt/Pt)	1980(4) -1987(4)	-9.264 (-4.280)***	1.146 (4.541)***	0.593 (3.444)***	-0.121 (-1.925)**	—	0.4789 (4.056)***	0.97	2.02	0.27 (0.895)	0.60 (0.665)	2.71 (5.911)

Source: Staff calculations.

^1/t-statistics in parentheses; one asterik indicates that the coefficient is significantly different from zero on the 10 percent level, two asterisks indicate significance on the 5 percent level and three asterisks on the 1 percent Level.

^2/Lagrange Multiplier Test for Auto Regression Conditional Heteroscedasticity; critical values at the 5 percent Level in parentheses.

^3/Lagrange Multiplier Test for Residual Autocorrelation; critical values at the 5 percent Level in parentheses; according to further tests the hypothesis of the non-existence of residual autocorrelation of different orders could not be rejected.

^4/Chi ² - Test for normal distribution of residuals; critical values on the 5 percent level in parenthesis

View Table

Table 2.

Regression Results ^1/

Dependent Variable	Period	Independent Variables						R2	Durbin-Watson	ARCH(4) 2/	$\mathrm{AR}{\left(4\right)}_1^4$3/	NORM(2) 4/
		Constant	$\mathrm{L}\mathrm{og}{\mathrm{Y}}_{\mathrm{t}}^{\mathrm{r}}$	it	π	Log (Ct-1/Pt-1)	Log (Dt-1/Pt-1)
Log (Ct/Pt)	1972(2) -1980(2)	-3.481 (-1.370)*	0.4195 (1.543)*	-0.775 (-1.507)*	-0.204 (-1.609)*	0.666 (5.960)***	—	0.75	2.05	0.28 (0.884)	0.124 (0.328)	1.138 (5.991)
Log (DtPt)	1972(2) -1980(2)	-7.509 (-3.171)***	1.069 (3.774)***	1.051 (2.289)**	-0.259 (-3.280)***		0.341 (2.583)**	0.84	1.89	0.58 (0.684)	0.0318 (0.0343)	1.123 (5.991)
Log (Ct/Pt)	1980(4) -1987(4)	-9.091 (-3.080)***	1.043 (3.186)***	—	-0.086 (-1.362)*	0.412 (2.618)**	—	0.86	1.95	0.51 (0.729)	0.124 (0.328)	0.59 (5.991)
Log (Dt/Pt)	1980(4) -1987(4)	-9.264 (-4.280)***	1.146 (4.541)***	0.593 (3.444)***	-0.121 (-1.925)**	—	0.4789 (4.056)***	0.97	2.02	0.27 (0.895)	0.60 (0.665)	2.71 (5.911)

Source: Staff calculations.

^1/t-statistics in parentheses; one asterik indicates that the coefficient is significantly different from zero on the 10 percent level, two asterisks indicate significance on the 5 percent level and three asterisks on the 1 percent Level.

^2/Lagrange Multiplier Test for Auto Regression Conditional Heteroscedasticity; critical values at the 5 percent Level in parentheses.

^3/Lagrange Multiplier Test for Residual Autocorrelation; critical values at the 5 percent Level in parentheses; according to further tests the hypothesis of the non-existence of residual autocorrelation of different orders could not be rejected.

^4/Chi ² - Test for normal distribution of residuals; critical values on the 5 percent level in parenthesis

As far as the demand for cash is concerned, all coefficients have the expected sign and are significantly different from zero at least at the 10 percent level. All the diagnostic tests are satisfied at a significance level of 5 percent; neither residual autocorrelation, nor non-normality nor heteroscedasticity can be detected. Similarly satisfying results were obtained with regard to the demand for deposits. Again, all coefficients show the expected sign while the t-values indicate a strong significance especially for real income and the inflation rate.

In order to evaluate the out-of-sample properties of the model, the estimations were used to forecast the demand for currency and deposits in the 1980s. The predicative accuracy was indeed poor. Supposing that the model has correctly been specified, then the predicative failure for the equations indicates that there has been a structural break in the demand for money. ^22/ In fact, the Chow-test for parameter constancy rejected the hypothesis of parameter stability. ^23/ The Chi² -test ^24/ was applied which confirmed the results of the Chow-test. As can be seen from Chart 3 the regression for the demand for currency implies a serious overestimation of the actual development in the indexation period, whereas the demand for deposits is heavily underestimated. This indicates that the introduction of monetary indexation has caused a significant shift in portfolio preferences.

Demand for Money: Actual and Fitted Values^1/

(In Logs)

Citation: IMF Working Papers 1990, 020; 10.5089/9781451921939.001.A001

Source: Staff calculations.^1/ In real terms (deflated by the GDP deflator)

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Demand for Money: Actual and Fitted Values^1/

(In Logs)

Citation: IMF Working Papers 1990, 020; 10.5089/9781451921939.001.A001

Source: Staff calculations.^1/ In real terms (deflated by the GDP deflator)

• Download Figure
• Download figure as PowerPoint slide

Demand for Money: Actual and Fitted Values^1/

(In Logs)

Citation: IMF Working Papers 1990, 020; 10.5089/9781451921939.001.A001

Source: Staff calculations.^1/ In real terms (deflated by the GDP deflator)

• Download Figure
• Download figure as PowerPoint slide

Re-estimating the model for the indexation period 1980(4) to 1987(4) produced the results also reported in Table 2. Almost all of the coefficients are highly significant and all of them have the correct sign. The diagnostic tests do not indicate that the regressions suffer residual autocorrelation, non-normality, and heteroscedasticity. Two points are particularly worth noting: first, the interest rate variable which was used for the estimation of the demand for money in the pre-indexation period proved to be no longer significant. However, as far as the demand for deposits is concerned, a strong influence of the indexed interest rate on time deposits could be detected. ^25/ In contrast, the opportunity costs of holding currency seem to be solely determined by the rate of inflation, whereas the return on alternative assets appears to play no longer a significant role. Experiments with different interest rates did not meet with success so that the interest variable was consequently omitted. Second, the (semi-) elasticity of the demand for money with respect to inflation has decreased significantly, both in the case of the demand for currency and for deposits. On the other hand, the elasticity with respect to real income has increased considerably, especially in the case of the demand for currency.

In the final part of the paper, the regression results obtained are used to estimate the demand for money and hence the revenues from money creation with and without indexation. This is done under alternative macroeconomic scenarios in order to evaluate different policy options.

VI. Some Policy Simulations

Because indexation of base money leads to lower revenues from money creation, the so-called financiable deficit, that is, the deficit which can be financed without jeopardizing target values for internal and external debt, becomes smaller. If the Government decides not to adjust fiscal policies and hence the actual deficit by cutting down expenditures or raising taxes, than those targets will not be attainable.

Since monetary indexation leads to a smaller base on which the inflation tax can be levied, one obvious possibility is to increase the tax rate, i.e., the inflation rate. In order to evalue this option in greater detail let us suppose that in the indexation scenario all required reserves are fully indexed, whereas in the nonindexation scenario required reserves do not even earn interest. These assumptions while somewhat unrealistic, help highlight the implications of indexing base money.

Under those assumptions a policy of raising inflation rates would have the following outcomes in 1985 (Table 3) ^26/

Table 3.

Revenues from Money Creation at Alternative Inflation Rates

(In percent of GDP)

Source: Staff calculations.

Table 3.

Revenues from Money Creation at Alternative Inflation Rates

(In percent of GDP)

Inflation Rate	Scenario I: Nonindexation	Scenario II: Indexation
			10	0.6	0.4
15	0.8	0.5
20	1.0	0.5
25	1.2	0.6
30	1.4	0.6
35	1.6	0.7
40	1.8	0.8
50	2.1	0.9
60	2.5	1.0
100	3.6	1.4

Source: Staff calculations.

View Table

Table 3.

Revenues from Money Creation at Alternative Inflation Rates

(In percent of GDP)

Inflation Rate	Scenario I: Nonindexation	Scenario II: Indexation
			10	0.6	0.4
15	0.8	0.5
20	1.0	0.5
25	1.2	0.6
30	1.4	0.6
35	1.6	0.7
40	1.8	0.8
50	2.1	0.9
60	2.5	1.0
100	3.6	1.4

Source: Staff calculations.

With money demand relatively inelastic to the opportunity cost of holding money, increased inflation generates increased revenues from monetization. ^27/ However, the marginal increase in revenues gradually falls as inflation rises because the demand for real balances falls. ^28/ Under indexation, an inflation rate of even 100 percent would cause revenues from money creation of only 1.4 percent of GDP. ^29/ In contrast, in the nonindexation scenario where base money is an interest-free liability of the Central Bank, revenues from money creation achieve this level at an inflation rate of only 30 percent. At an inflation rate of 100 percent total revenues are more than 2 1/2 times larger than in the case of indexation. This difference would of course be even larger if required reserves received a positive real rate of interest as it has actually been the case since March 1988.

Revenues from money creation are only one source of financing. In addition there is domestic debt issue and foreign borrowing ^30/ In order to estimate the financiable deficit, all three sources have to be combined. Thus calculated, the financiable deficit can then be compassed with the actual deficit, with the difference between the two measures indicating the required deficit reduction (RDR). ^31/ This has been done in Table 4 whereby the analysis focuses not only on changes in the inflation rate, but also on changes in economic growth and the real exchange rate.

Table 4.

Required Deficit Reductions for Consistency with Various Macroeconomic Targets

(In percent of GDP)

Source: Staff calculations.

Table 4.

Required Deficit Reductions for Consistency with Various Macroeconomic Targets

(In percent of GDP)

		No Real Depreciation, Real GDP Growth of 3 Percent
Inflation rate		10	20	30
RDR
	Nonindexation	0.1	-0.3	-0.7
	Indexation	0.4	0.3	0.2

		No Real Depreciation, Inflation Rate of 10 Percent Inflation Rate
Growth target		2	3	4
RDR
	Nonindexation	0.8	0.1	-0.5
	Indexation	1.1	0.4	-0.3

		Real GDP Growth of 3 Percent, Rate of 10 Percent Inflation
Real depreciation target		—	2	4
RDR
	Nonindexation	0.1	0.9	1.8
	Indexation	0.4	1.3	2.2

Source: Staff calculations.

View Table

Table 4.

Required Deficit Reductions for Consistency with Various Macroeconomic Targets

(In percent of GDP)

		No Real Depreciation, Real GDP Growth of 3 Percent
Inflation rate		10	20	30
RDR
	Nonindexation	0.1	-0.3	-0.7
	Indexation	0.4	0.3	0.2

		No Real Depreciation, Inflation Rate of 10 Percent Inflation Rate
Growth target		2	3	4
RDR
	Nonindexation	0.8	0.1	-0.5
	Indexation	1.1	0.4	-0.3

		Real GDP Growth of 3 Percent, Rate of 10 Percent Inflation
Real depreciation target		—	2	4
RDR
	Nonindexation	0.1	0.9	1.8
	Indexation	0.4	1.3	2.2

Source: Staff calculations.

As Table 4 shows, required deficit reductions are considerably higher in the case of indexation. Assuming that real economic growth is 3 percent and that the real exchange rate does not depreciate, fiscal adjustment would still be necessary at an inflation rate of even 30 percent. In contrast, if required reserves are not indexed, the financiable deficit would be equal to the actual deficit at an inflation rate of only slightly more than 10 percent. However, if economic growth accelerates, the need for fiscal adjustment decreases. At a growth rate of 4 percent the financiable deficit exceeds the actual deficit even in the case of indexation (under the assumption that the inflation rate is only 10 percent.) In turn, the need for fiscal adjustment increases, if economic grwoth decelerates, whereby the required deficit reduction is relatively greater in the case of indexation. ^32/

For the Icelandic economy, which is subject to severe shocks from its heavy reliance on fisheries, these results are of great interest. Suppose, for example, that as a result of an exogenous shock there is a decline in the rate of economic growth. Such a decline will, ceteris paribus, necessitate a compensating increase in the steady-state rate of inflation in order to keep the Government’s revenues from money creation unaffected. The required increase in the rate of inflation is however substantially larger in the case of indexation than with non-indexation. As it has been shown by Melnick and Sokoler (1984), a decline in the growth rate may not only call for an increase in the rate of inflation (an increase greater than the decline in the rate of growth), but it may also imply that the necessary compensating increase in the rate of inflation is a positive function of the existing rate of inflation. This will be the case if the demand for real balances is a semilogarithmic function of the expected rate of inflation, or in other words, if the elasticity of the demand for money with respect to the expected rate of inflation increases as the rate of inflation rises. ^33/

In order to prevent a deceleration of economic growth, the authorities might try to pursue an export strategy that relies on real exchange rate depreciation. However, such a policy would imply a greater need for fiscal adjustment by raising the ratio of foreign debt to GDP. Because of Iceland’s high external debt, the required deficit reduction increases remarkably with the rate of real depreciation. Assuming real economic growth of 3 percent and an inflation rate of 10 percent, real depreciation of 4 percent would cause a gap of 1.8 percent of GDP between the financiable deficit and the actual deficit even in the case of nonindexation. If required reserves are fully indexed, the necessary fiscal adjustment would be even larger.

Further simulations are conceivable, for example, to explore the consequences of changes in reserve requirements ^34/ or in bank interest rates. ^35/ However, it should already have become sufficiently clear that monetary indexation has important budgetary implications.

VII. Conclusions

As has been shown in this paper, monetary indexation has significant budgetary consequences which are all the more important for countries like Iceland, where revenues from money creation represent a substantial part of total Government revenues. Starting from the observation that revenues from money creation declined sharply after indexation was introduced, a simple framework has been presented in which the consistency of fiscal deficits with other macroeconomic targets could be assessed. This framework based on the public finance approach to inflation allowed us to consider the relations of fiscal deficits to other targets including output growth, real exchange rate developments and internal and external debt with and without indexation. By simulating different policy options it can be shown that, ceteris paribus, monetary indexation considerably raised the need for fiscal adjustment.

This would not have been the case, if only financial assets in the banking system were indexed. On the contrary, due to the increase in seignorage associated with larger bank deposits and required reserves, required deficit reductions, would have been comparatively lower. However, in the absence of base money indexation the spread between deposit and lending rates would probably increase significantly.

In the recent past, steps have been taken to reduce the extent of monetary indexation. As an alternative to linking monetary assets to the price development, deposits have been introduced where assets are linked to the exchange rate. These schemes may provide not only an alternative to deposits indexed to the credit terms index but also an alternative to foreign exchange deposits. While foreign exchange linked deposits do not require the use of actual foreign currency, foreign exchange deposits normally lower the base on which the inflation tax can be levied -- insofar as they provide a liquid alternative to domestic money. ^36/ However, in Iceland foreign currency deposits are nevertheless subject to reserve requirements so that this potential effect is of no importance. ^37/