A. Fiscal policy

Prior to the beginning of the stabilization period, Armenia had an expansionary fiscal policy characterized by extremely high budget deficits financed predominantly by CBA credit. Over the course of adjustment, it moved to a tighter fiscal stance resulting in substantially reduced budget deficits. The accrual deficit of the consolidated government budget fell by 40 percentage points of GDP in 1994 to 16.5 percent of GDP, with continued declines to 9.9 and 8.8 percent of GDP in 1995 and 1996, respectively. The composition of budgetary financing also changed significantly. While in 1993 financing came mainly from the CBA and from external sources (34 and 21 percent of GDP, respectively), financing in 1994 consisted mainly of domestic expenditure arrears and external financing (about 6 percent of GDP each), while central bank financing fell to 3.5 percent of GDP.

Dramatic changes occurred on the expenditure side, with total expenditure plummeting from 85 percent of GDP in 1993 to 44 percent in 1994, falling further to 30 percent in 1995 and 26 percent in 1996, as a result of massive across-the-board reductions in spending. In 1996, the share of current expenditures in GDP stood at one-third its 1993 level; during the same period, the share of capital expenditures and net lending in GDP declined to one-fourth its 1993 level. Capital expenditure fluctuated throughout the stabilization period, reaching almost 10 percent of GDP in 1994 but declining to 7 percent of GDP in 1995 and then to below sustainable levels at around 4 percent of GDP in 1996. Net lending to enterprises--an implicit subsidy to the enterprise sector, given highly negative real interest rates and poor repayment discipline-- fell from 18 percent of GDP in 1993 to negligible levels in 1994-95, and about 2 percent of GDP in 1996. In sum, the bulk of fiscal adjustment occurred on the expenditure side, especially in 1994, when inefficient spending was drastically reduced, including some reorientation of expenditures from defense, given improving prospects for peace in the region.⁷

Reductions in expenditure were accompanied by sizable drops in revenues and grants, particularly of the latter. Following a peak of around 12 percent of GDP in 1994, foreign grants accruing to the budget (mainly the domestic counterpart to food aid in the form of flour and grain used for bread) dropped to under 4 percent of GDP in 1995, and remained at that level in 1996. Tax revenues declined from 16 percent of GDP in 1993 to around 13 percent throughout the stabilization period due to administrative difficulties in taxing newly emerging sectors of the economy, the lower level of profitability in the state enterprise sector, persisting tax arrears, and weak performance of the VAT mainly related to widespread exemptions. Despite efforts to broaden the tax base, the state sector remained by far the largest source of tax revenues, accounting for about 90 percent of the total throughout the stabilization period. However, this implied a steadily declining tax base due to privatization and the increasing share of the underground economy. These forces, coupled with a drastic reduction in the inflation tax (see section III) overwhelmed gains from the reverse Tanzi effect as the level of inflation fell. Nontax revenues declined from 8 percent of GDP in 1993 to around 3 percent in part due to lower interest rates, which resulted both in lower interest income on net lending by the budget, and lower profit transfers from the CBA.

During the stabilization period, two instances of temporary fiscal loosening occurred. The increase in the budget deficit at the end of 1995 was mainly due to a surge in tax arrears, and the failure to lower expenditures in the face of weak revenue performance; the fiscal expansion during the third quarter of 1996 was driven by sharp increases in spending on goods and services and government lending to enterprises mainly financed through CBA credit to government, clearly related to the end-September presidential elections. In both cases the subsequent tightening constituted a return to the earlier stance.

Throughout the stabilization period, tax arrears persisted, and the budget repeatedly incurred expenditure arrears as well. One way the Ministry of Finance attempted to deal with this problem was through netting-out operations, essentially involving the use of promissory notes (IOUs issued by budgetary spending units to enterprises as payment for goods and services bought from them) to pay back tax arrears. These netting-out operations were recorded in the fiscal accounts on both the expenditure side (as expenditure by the spending unit) and the revenue side (as tax revenue when the tax arrears were cleared).

However, these operations posed a number of problems. They gave rise to moral hazard (a clear incentive was created to run up tax arrears, which could be repaid later, and often at a discount in real terms) and were a non-transparent means of dealing with the problem of tax arrears, providing a misleading picture of the underlying revenue performance. Given the promissory notes’ degree of substitutability with T-bills, and the active secondary market in promissory notes at the Yerevan stock exchange that developed by early 1996, their existence complicated plans for expanding the T-bill market and the development of an orderly secondary market for them; these notes also added to liquidity in the economy which was outside the control of the CBA. Their use was discontinued from August 1996.

The general trend toward a declining budget deficit during the stabilization period does not necessarily imply that the fiscal stance was uniformly contractionary in 1994-96, or that the revenue and expenditure stance was similar. An analysis of the fiscal stance during the stabilization period based on fiscal impulse calculations is presented in Section III. These indicators were chosen over changes in structural balances which are also used in the literature, because they are parameter-free, i.e. they measure the policy stance without recourse to elasticity estimates.

Section III also provides measures of seigniorage and the inflation tax extracted from the economy and its evolution during 1994-96, which show that inflationary financing of the budget has all but disappeared. As fiscal policy was tightened and the monetary stance became more restrained during 1994-1995, the rapid decline in inflation had the side-effect of sharply curtailing seigniorage and the inflation tax, hitherto very important sources of financing the general government. On an annual basis, general fiscal financing through inflation as measured by seigniorage dropped by a factor of six in 1995, despite a substantial decline in measured dollarization; a further similar decline occurred in 1996. The magnitude of the decline in the inflation tax is comparable over the same period, reinforcing the view that the successful stabilization of the Armenian economy during 1994-95 reduced budgetary financing through inflation to negligible levels.

The primary direction of causality during the stabilization period was from financial tightening to lower inflation. However, as fiscal consolidation contributed to bringing about a sharp drop in inflation, it led to lower indirect budgetary revenues arising as a result of inflation.⁸ Tax revenue also fell, as described above. As a result, the sum of budgetary tax revenue and the inflation tax declined in a dramatic manner--from over 60 percent of GDP in the first quarter of 1994 to under 15 percent in 1996--making further fiscal adjustment inevitable. Thus, the government was forced to proceed with the drastic expenditure cuts that were implemented during the stabilization period. Despite their adverse impact, these developments also had beneficial effects: the sharp reduction in the proportion of GDP channeled through the government budget had a crowding in effect on private sector activity, most notably in the expanding service sector, which underpinned growth throughout the stabilization period. In addition, the substantial decline in the effective tax burden on the economy also facilitated growth.

Pursuing an alternative strategy of allowing a markedly larger fiscal deficit and higher inflation in order to achieve faster growth and increased government financing would have been excessively risky. In particular, little of the past experience with high deficits and expanding credit provided hope that they could improve allocational efficiency. Moreover, volatile exchange rate expectations had to be addressed through decisive policy action, since the emergence of macroeconomic stability owed much to the effectiveness of the de facto nominal exchange rate anchor (see below). Thus, on balance, the tightening of fiscal policy played a positive role by helping to move the economy toward an equilibrium with low inflation and a stable exchange rate, and placing it on a sustained growth path.

B. Monetary policy

Monetary policy in Armenia was successful during the stabilization period, as demonstrated by the attainment of ambitious inflation and output objectives. Since these variables were not directly controllable and also because of the lags in the effect of financial policies on prices and output, the CBA focused on intermediate targets, including the exchange rate, NIR, and bank liquidity. These were more immediately controllable, with data available on a timely basis. In addition, exchange rate developments were widely followed as a highly visible indicator of policy commitment. The CBA had a limited set of instruments to attain desired levels for its intermediate targets. While the CBA argued that large disbursements of official loans accruing to the government should be kept in the government account at the CBA and spent (i.e. released into the banking system) only gradually in order to avoid a surge in dram liquidity, in practice such disbursements were always promptly spent. The main--in fact, until 1996, practically the only--instrument for withdrawing substantial liquidity in these circumstances was the massive sales of foreign exchange. These sales took place in different segments of the foreign exchange market in different periods: in the regular foreign exchange auctions during 1994-95, when these auctions were the main determinants of the exchange rate; and in the emerging interbank market subsequently, when it took over this role from mid-1996. The CBA was an important--often dominant--participant in the foreign exchange market throughout the stabilization period, balancing the conflicting objectives of limiting the liquidity impact of lump-sum inflows of official external grants and loans, and maintaining a stable exchange rate, without formally fixing it. Given the relatively low level of reserves and the volatile external environment, the CBA viewed this approach as preferable to a formal fixing of the exchange rate, which could have proven costly in terms of credibility if unforeseen shocks forced an abandonment of the fixed rate.

The CBA‘s primary instrument for injecting dram liquidity was the provision of central bank credit to banks. Following a period when direct allocation of CBA credit to banks prevailed in 1994, the procedures for central bank credit auctions were overhauled in February 1995 and these gradually became the main means of managing liquidity⁹ through end-1995. However, following the introduction in September 1995 of Treasury bill auctions, the CBA decided to impose a collateral requirement on all auctioned credit from January 1 1996 in order to reduce the repayment risk of auctioned credit and to facilitate the emergence of an interbank credit market. Unfortunately, the immediate effect of this measure was to effectively eliminate credit auctions as an instrument from that time due to the insufficient stock of outstanding T-bills available for this purpose,¹⁰ with the resulting credit crunch being felt throughout the first half of 1996. The liquidity impact was partly offset by the effective lowering of the reserve requirement in May 1996; in the event, the use of this blunt and rigid instrument for short term liquidity management proved inefficient, and was subsequently reversed. The CBA also introduced repurchase agreements (repos), reverse repos and a Lombard facility in early 1996; however, their role remained insignificant through the end of the year.

While the interbank market in dram credit registered noticeable gains from the second half of 1996, it did not open up significant new transmission mechanisms for monetary policy during the stabilization period (e.g. through interest rate policy). The CBA set the refinance rate, but it was not used in any significant transaction in the economy (see the discussion of advance profit transfers by the CBA). The decision to auction the right to banks to place remunerated deposits at the CBA from August 1996 may over time increase the role of interest rates in the economy. However, auctioned deposits remained a marginal instrument in 1996, in part due to the desire not to interfere excessively with the primary market in short term T-bills, for which they are a close substitute. The interbank credit market lacked sufficient depth and remained segmented due to the high perceived payment risk of many banks, including major ones, and to the less-than-transparent collateral legislation. An additional problem was that no secondary market in T-bills emerged, due to the discriminatory taxation of secondary transactions in T-bills, and their short average maturity. Such a secondary market could have facilitated a faster expansion of non-inflationary budget financing through T-bills, enhanced liquidity management by the CBA and a more rapid development of the interbank credit market.

An important leakage in restricting CBA financing to the budget was the system of advance profit transfers, whereby until end-1996 the CBA automatically covered the quarterly payments of interest due on CBA credit by a transfer to the budget which was referred to as advance profit transfers. These advance transfers were substantial in 1994 (around 8 percent of GDP), and declined substantially by 1996. It is noteworthy that they are roughly of the same magnitude as the calculated amounts of quarterly seigniorage accruing to the CBA during 1994-96 (see Section III).

By late 1993, Armenia had no significant monetary overhang due to the high level of inflation in the preceding period.¹¹ Thus, the subsequent monetary tightening played a major and immediate role in lowering inflation during the initial stabilization phase. The rate of growth of both base and broad money was under one-thirtieth in 1995 of what they were in 1993 (Table 1), and this fundamental shift toward tighter monetary policy coincided with a sharp drop in inflation.¹² However, from late 1995, monetary policy exhibited features of a stop-go cycle. Serious monetary policy relaxations occurred in the fourth quarter of 1995 and the third quarter of 1996, driven primarily by the fiscal lapses described above. The fact that inflation did not flare up was due to the prompt correction applied in both instances, and other, specific factors. At end-1995, the sudden increase in NIR which pushed up base money was rapidly reversed; in addition, there was an increase in the demand for drams as suggested by the downward drift in the share of foreign exchange deposits in broad money from a peak of 63 percent in May 1994 to 20 percent by the fourth quarter of 1995 (Figure 4). In the second case, the effect of the surge in monetary aggregates on measured inflation fed through with an unexpectedly long lag (inflation did pick up during 1997). In addition, inflationary pressures may have partly found an outlet on the external side: the current account deficit, although declining, remained high as a percentage of GDP, while the nominal exchange rate depreciated substantially during the last quarter of the year. The stop-go cycle in financial policies did not represent an optimal response to changing domestic macroeconomic circumstances and may have complicated the task of permanently altering the expectations of economic agents by exacting a toll on policy credibility.

The interaction between fiscal and monetary policy has also evolved over time. Until 1993, the CBA just financed the fiscal deficit that emerged, without much regard to inflationary consequences. From 1994, it began sharply curtailing its financing to the budget, and over time bank financing was also reduced. However, this component remained substantial due to the fact that banks were the main purchasers of T-bills, in effect providing bank financing to the budget. As the description of monetary policy instruments shows, monetary stability throughout the stabilization period hinged largely on continued fiscal restraint and the success of sterilization operations through sales of foreign exchange. It was therefore not surprising that the sharp increase in credit to government related to upcoming elections in September 1996 was followed by a substantial loss of CBA reserves, and subsequently, by exchange rate depreciation. This episode of loose financial policies was captured by both impulse indicators presented in section III.

After careful consideration the authorities did not formally commit to a fixed exchange rate system at the outset of the stabilization process. However, the exchange rate vis-a-vis the US dollar remained practically constant without substantial CBA intervention for almost two years through mid-1996, facilitated by the dram’s initial undervaluation in early 1994. This in turn was due to (i) the dram’s sharp initial depreciation in the months after its introduction; and (ii) the reversal of the initially high demand for foreign exchange which had been regarded as a superior store of value before stabilization took hold, given the negative real interest rates and high risk on domestic savings instruments. The resulting de facto nominal exchange rate anchor has played an important role in lowering inflation. As the public’s confidence in the dram grew, reverse currency substitution occurred, driving up the demand for drams and facilitating the stability of the nominal exchange rate, causing a strong real appreciation of the dram during much of the stabilization period due to the significant inflation differential with most partner countries. The stable nominal exchange rate helped limit inflation both through its effect on expectations, and by effectively limiting the rate of price increases of imported goods to foreign inflation. The latter effect was strengthened by the gradual reduction in traded goods prices due to increasing price arbitrage as the trade blockade became more porous over time. The stability of the exchange rate also provided a highly visible signal of the overall appropriate stance of financial policies and thus helped guide expectations towards a low inflation equilibrium.

A. Fiscal Impulse Indicators

It has long been recognized that for many countries the actual budgetary position of the government may be a misleading indicator of the thrust of fiscal policy, because it is not clear whether changes in that position are the cause or the result of changes in economic activity (cf. Heller et al, 1986). In addition, when analyzing fiscal policy it is important to distinguish between factors that have a transitory effect on the actual budget balance, and the effects of policy or structural changes that have a more durable impact on the budget balance. These concerns, which also apply to the Armenian economy during the period analyzed, led to the development of the fiscal impulse measure for adjusting the fiscal accounts to yield a more accurate measure of the fiscal policy stance. This section presents such quantitative fiscal impulse indicators for Armenia, which characterize the effect of fiscal policy on overall aggregate demand in the economy during 1994-96.

The quantified measures can be interpreted as follows (technical details are presented in Appendix I). The revenue impulse considers the impact of government revenues neutral if revenues and actual GDP grow proportionally; if revenues grow more than proportionally the effect is contractionary, and if they grow less than proportionally, the impact is expansionary. On the other hand, the expenditure impulse considers the impact of government expenditures on the economy neutral if they grow proportionally to potential output. If expenditures grow more than proportionally to potential output, the impact is considered expansionary, and if they grow less than proportionally, the impact is contractionary. The fiscal impulse, i.e., the sum of the revenue and the expenditure impulses, measures the net effect of the revenue and expenditure policies of the government on aggregate demand. Both annual and quarterly impulse indicators were calculated, although the emphasis in the analysis is on the annual ones. For the quarterly impulse indicators, year-on-year differences were used so that the quarterly impulse indicators add up to the annual indicators. Total revenues include tax and nontax revenues and humanitarian assistance counterpart funds accruing to the budget,¹⁴ while government expenditures include net lending.

The fourth quarter of 1995 was chosen as the base period for the expenditure impulse measures described above. The first implication of this choice is that the revenue/GDP and expenditure/GDP ratios of that quarter are used to define the cyclically neutral budget (see Appendix I). Thus, the fiscal stance in each period is compared to the stance of this quarter. It is important to keep in mind that fiscal policy in period “t” can be expansionary—compared to the base period—even if strong fiscal adjustment occurred, i.e., the deficit declined sharply compared to period “t-1”. As Appendix I indicates, the sign of the revenue impulse is determined not only by the change of revenue in relation to the change of GDP, but also by its relative magnitude compared to the revenue/GDP ratio at the base period, t0. It is easy to show that the revenue impulse would be positive (expansionary), if the revenue/GDP ratio declined from period “t-1” to period “t” to a level below t0; or alternatively, negative (contractionary), if the revenue/GDP ratio increased from period “t-1” to period “t” to a level above to. A similar analysis can be applied to determine the sign of the expenditure impulse. In sum, the choice of the base ratios affect the size, and possibly even the sign of the fiscal impulse indicators. However, their relative magnitude is essentially unaffected: periods with a high or low impulse compared to other periods remain such even if different base ratios are selected.

The second implication of the choice of the fourth quarter of 1995 as the base period is that the level of real GDP in this quarter was used as a measure of potential GDP.¹⁵ In the absence of capacity measures and lacking a general equilibrium model for the economy, this choice is necessarily arbitrary. It is based on the local peak in output achieved in this quarter, which appeared to exceed the level implied by seasonality; the fact that this was the last quarter for which Fund programs aimed primarily at stabilization (STF, SBA) were in effect; and that high levels of inflation, together with rapidly changing relative prices ruled out quarters at the beginning of the sample period as suitable candidates. In order to ensure that this choice for potential GDP did not unduly distort the results, a sensitivity analysis was performed to check for its impact. This analysis is also presented in Appendix I, and essentially implies that the major conclusions of the exercise are robust to the choice of the period providing the level of potential GDP.

Turning to the results, it is readily apparent that during the period analyzed, the expenditure impulse dominated the revenue impulse, and hence determined the overall stance of fiscal policy. This was particularly noteworthy in 1994, when despite considerable contractionary effects on the revenue side, the overall fiscal impulse shows a marginally expansionary stance due to a strongly expansionary expenditure effect. The contractionary revenue impulse was a result of the sharp slow-down in inflation, the introduction or amendment of a number of new taxes including VAT and land taxes, and significant reductions in exemptions. The major reasons for the strong expansionary expenditure effect were the high level of subsidies—12.8 percent of GDP, compared to 0.6 percent in Q4 1995, the base period, and of capital expenditures—9.8 percent of GDP, compared to 6.4 percent in Q4 1995.

As Table 2 and Figure 1 show, the overall fiscal stance in all but two quarters arose as a result of revenue and expenditure impulses of opposite signs, reflecting an uneven approach to fiscal policy. Following a strong expansion in 1994, drastic expenditure cuts dominated the scene in 1995-96, especially in the subsidies, transfers, and other goods and services items, reducing expenditures to levels from which further reductions of similar magnitude were no longer feasible. The path of the revenue impulse is the mirror image: while it held the overall fiscal impulse at a relatively low level in 1994, it showed an expansionary stance (i.e., weaker revenue effort) in 1995-96. Overall, the fiscal impulse results characterize fiscal policy as having been somewhat expansionary in 1994, with a reversal already apparent in the final quarter, strongly contractionary in 1995 and mildly contractionary in 1996. Of the two components of the fiscal impulse, the expenditure impulse dominated the revenue impulse in magnitude in all three years. The same holds for the quarterly measures, with the exception of the third quarter of 1996. In sum, fiscal policies have been substantially tightened over the period analyzed, primarily on the revenue side in 1994 and predominantly the expenditure side in 1995-96.

View Full Size

Armenia: Fiscal Impulse, 1994-96

(In percent of GDP)

Citation: IMF Working Papers 1998, 038; 10.5089/9781451845662.001.A001

sources: Armenian authorities; and Fund staff estimates.

• Download Figure
• Download figure as PowerPoint slide

Table 2.

Armenia: Indicators of Fiscal Impulse ^1/

(In percent of GDP)

^1/A positive (negative) quantity denotes an expansionary (contractionary) impact.

Table 2.

Armenia: Indicators of Fiscal Impulse ^1/

(In percent of GDP)

Time	Fiscal Impulse	Revenue Impulse	Expenditure Impulse
1994	1.79	-6.64	8.44
Q1	0.59	-0.91	1.49
Q2	0.44	-3.44	3.88
Q3	2.40	-2.01	4.41
Q4	-1.63	-0.28	-1.35
1995	-4.53	3.36	-7.90
Q1	-1.72	2.04	-3.75
Q2	-0.40	0.73	-1.13
Q3	-1.83	0.49	-2.32
Q4	-0.59	0.11	-0.70
1996	-1.70	3.82	-5.52
Ql	-0.13	0.34	-0.47
Q2	-0.56	1.53	-2.09
Q3	0.97	2.41	-1.44
Q4	-1.97	-0.45	-1.52

^1/A positive (negative) quantity denotes an expansionary (contractionary) impact.

View Table

Table 2.

Armenia: Indicators of Fiscal Impulse ^1/

(In percent of GDP)

Time	Fiscal Impulse	Revenue Impulse	Expenditure Impulse
1994	1.79	-6.64	8.44
Q1	0.59	-0.91	1.49
Q2	0.44	-3.44	3.88
Q3	2.40	-2.01	4.41
Q4	-1.63	-0.28	-1.35
1995	-4.53	3.36	-7.90
Q1	-1.72	2.04	-3.75
Q2	-0.40	0.73	-1.13
Q3	-1.83	0.49	-2.32
Q4	-0.59	0.11	-0.70
1996	-1.70	3.82	-5.52
Ql	-0.13	0.34	-0.47
Q2	-0.56	1.53	-2.09
Q3	0.97	2.41	-1.44
Q4	-1.97	-0.45	-1.52

^1/A positive (negative) quantity denotes an expansionary (contractionary) impact.

As prices were liberalized and subsidies fell, expenditure policy became contractionary, and remained consistently so throughout 1995 with total expenditures declining from 44.1 percent of GDP in 1994 to 29.8 percent, leading to an overall contractionary fiscal stance in that year. This occurred despite a revenue policy that was expansionary, especially in the first quarter of 1995 due to a drop in tax revenues to only 9.3 percent of GDP from 15 percent in the preceding quarter. By end-1995, excise tax measures reversed the decline in tax revenues and thus made the end-year revenue impulse close to neutral. Fiscal policy was also uneven in 1996: it was essentially neutral during the first half, followed by a strongly expansionary stance in the third quarter¹⁶ and a reversal to contraction in the final quarter. The reversal was facilitated by a sharp increase in tax arrears payments following a presidential decree threatening state enterprise managers with firing if tax arrears were not paid by end-1996.

B. Seigniorage and Inflation Tax

Armenia experienced very high rates of base money expansion and inflation in 1993 and early 1994. This resulted in high levels of seigniorage¹⁷ and inflation tax, which are quantified using a simple approach in this section, based on the methodology in Buiter (1996). For the purposes of this paper, seigniorage is defined as the amount of resources appropriated during a period by the government¹⁸ as a result of expanding nominal base money:

where M_t is nominal base money in dram at end of period t and Y_t is nominal GDP in dram in period t. The inflation tax is defined as the reduction in the real value of the outstanding stock of base money due to inflation over a specified period:

where P_t is end-period consumer price base index for period t. Both seigniorage and inflation tax are measured as a percentage of GDP, and provide measures of the magnitude of resources appropriated by the government as a result of expanding the public’s holding of base money, and of inflation, respectively. They will differ if there is real growth in GDP, if real base money expands (the former raises seigniorage relative to the inflation tax, while the latter has the opposite effect, ceteris paribus), or if base money velocity changes.

The results (Table 3 and Figure 2) highlight an important aspect of stabilization: following very high levels, both seigniorage and the inflation tax declined to low, at times even negative, levels by mid-1995, as stabilization took hold in the economy. However, the results should be interpreted as broad indicators of trends only, due to the limitations of the data base and the lack of adjustment for possible changes in demand for base money. For the fourth quarter of 1993, data inconsistencies precluded the calculation of a reliable inflation tax figure, while seigniorage is omitted because it accrued to the Central Bank of Russia during part of this period before the dram was introduced.

View Full Size

Armenia: Seigniorage and Inflation Tax, 1994-96

(In percent of GDP)

Citation: IMF Working Papers 1998, 038; 10.5089/9781451845662.001.A001

Sources: Armenian authorities; and Fund staff estimates.

• Download Figure
• Download figure as PowerPoint slide

Table 3.

Armenia: Seigniorage and Inflation Tax

Source: Armenian authorities; and Fund staff estimates.

Table 3.

Armenia: Seigniorage and Inflation Tax

	CPI	Seigniorage	Inflation Tax
	1994 Jan=100	(in percent of GDP)	(in percent of GDP)
1993
Q4	55	…	…
1994
Q1	171	43.5	51.2
Q2	489	14.5	36.4
Q3	530	4.9	1.5
Q4	1,087	4.3	9.1
1995
Q1	1,152	6.7	0.7
Q2	1,338	-0.3	3.0
Q3	1,248	0.8	-1.2
Q4	1,435	3.9	2.4
1996
Q1	1,540	-1.3	1.5
Q2	1,514	1.3	-0.3
Q3	1,453	4.2	-0.7
Q4	1,518	2.4	0.9

Source: Armenian authorities; and Fund staff estimates.

View Table

Table 3.

Armenia: Seigniorage and Inflation Tax

	CPI	Seigniorage	Inflation Tax
	1994 Jan=100	(in percent of GDP)	(in percent of GDP)
1993
Q4	55	…	…
1994
Q1	171	43.5	51.2
Q2	489	14.5	36.4
Q3	530	4.9	1.5
Q4	1,087	4.3	9.1
1995
Q1	1,152	6.7	0.7
Q2	1,338	-0.3	3.0
Q3	1,248	0.8	-1.2
Q4	1,435	3.9	2.4
1996
Q1	1,540	-1.3	1.5
Q2	1,514	1.3	-0.3
Q3	1,453	4.2	-0.7
Q4	1,518	2.4	0.9

Source: Armenian authorities; and Fund staff estimates.

As in the case of the fiscal impulse, and the monetary impulse presented below, the seigniorage measures are relatively large in the final quarter of 1995 and the third quarter of 1996. However, this reflects the side-effects of financial loosening, rather than policy objectives. The short term respite brought about by the temporary increase in seigniorage was likely to have been offset by the damage to the credibility of future policies caused by the policy reversals. Moreover, while both the seigniorage and inflation tax measures indicated a greater extraction of resources from the economy at the end of 1995, this was only the case for seigniorage in the third quarter or 1996, reflecting the continued subdued level of inflation during this period.

C. Monetary Impulse Indicators

The monetary impulse measures presented here seek to capture the medium term inflationary implications of the stance of monetary policy in a simple measure. The main macroeconomic variables used are GDP and base money, the latter because it closely reflects actions taken by the monetary authority itself. Since the multiplier has been relatively stable during the sample period, the behavior of broad money was rather similar to that of base money. Thus, broad money-based calculations would have not yielded qualitatively different results for the period analyzed.

Following the approach in McCallum and Hargraves (1996), three different monetary impulse indicators were calculated. They are defined as quarterly rates of base money growth adjusted by three alternative measures of smoothed base money velocity (see Appendix II). They closely resemble a rearranging of the identity defining base money velocity, with the growth rate of GDP on the left hand side. Therefore, they represent growth rates of nominal GDP implied by the current stance of monetary policy as measured by changes in base money. Assuming that monetary policy should aim for very low inflation, these measures should be contrasted with levels of nominal GDP growth with approximate price stability. In order to account for (i) uncertainty about the precise level to be chosen, (ii) feedback effects of GDP growth on changes in base money, and (iii) the potential lack of stability in the demand for dram, a range, rather than a single level of stable-price GDP growth is chosen to represent a neutral monetary stance.¹⁹

To determine this range, a 5 percent annual real growth rate was chosen as a reasonable growth rate to aim for during the stabilization period, which followed a sharp drop in output. Approximate price stability in the Armenian context was taken to mean 6 percent annual inflation—corresponding to the actual 1996 level of end-period inflation in Armenia. These figures imply a nominal GDP growth rate of around 11 percent. Based on these considerations, a range of 3 percentage points above and below 11 percent was taken to be the range of noninflationary growth rates for nominal GDP.

Against this background, the results can be interpreted as follows. If the monetary impulse measures for a given quarter tend to fall within the 8-14 percent range described above, the monetary stance is considered neutral in that period; if above, it is considered expansionary, and if below, contractionary.

The limitations of these measures include the failure to capture changes in policy instruments and macroeconomic variables that act as substitutes for domestic credit, such as interenterprise arrears, informal credits extended outside the banking system and promissory notes. The monetary impulse measures are less likely to reflect the true monetary stance if progress on the stabilization front leads to autonomous capital inflows reflecting increases in the demand for dram.²⁰ In addition, to the extent that a central bank targets the exchange rate during 1994-96, the monetary impulse does not measure the intent of monetary policy, although it still provides a specific quantitative measure of its effect. Finally, changes in data definitions (including in the method of compiling GDP statistics, and their coverage), in the regulatory system, and structural changes in the banking system may have made data from different time periods less comparable.

The quantitative results are summarized in Figure 3 and Table 4 below. The interpretation is based on the overall picture emerging from the three monetary impulse indicators. While they yield somewhat different results for specific quarters, they are sufficiently similar to draw the following conclusions.

View Full Size

Armenia: Quarterly Monetary Impulse Measures and Inflation, 1994-96

(In percent)

Citation: IMF Working Papers 1998, 038; 10.5089/9781451845662.001.A001

Sources: Armenia authorities; and Fund staff estimates.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Armenia: Foreign Currency Deposits and Exchange Rate, 1994-96

Citation: IMF Working Papers 1998, 038; 10.5089/9781451845662.001.A001

Sources: Central Bank of Armenia; and Fund staff estimates.

• Download Figure
• Download figure as PowerPoint slide

Table 4.

Armenia: Monetary Impulse Measures ^1/

(in percent)

^1/See Appendix II for definitions. As indicated above, a measure exceeding 14 percent is considered expansionary; below 8 Percent, contractionary; and in between, neutral.

Table 4.

Armenia: Monetary Impulse Measures ^1/

(in percent)

	Backward-lookine MA		Centered MA
	No lag MI1	1-quarter lag MI2	MI3
1994
Q2	91.9	…	55.5
Q3	77.9	60.1	61.3
Q4	89.4	94.9	96.0
1995
Q1	53.8	92.4	63.7
Q2	-19.5	8.1	5.5
Q3	6.1	-13.4	-7.0
Q4	17.9	23.7	20.0
1996
Q1	-8.5	-10.4	-7.3
Q2	7.1	4.5	3.8
Q3	28.0	32.6	31.1
Q4	5.9	7.3	6.3

^1/See Appendix II for definitions. As indicated above, a measure exceeding 14 percent is considered expansionary; below 8 Percent, contractionary; and in between, neutral.

View Table

Table 4.

Armenia: Monetary Impulse Measures ^1/

(in percent)

	Backward-lookine MA		Centered MA
	No lag MI1	1-quarter lag MI2	MI3
1994
Q2	91.9	…	55.5
Q3	77.9	60.1	61.3
Q4	89.4	94.9	96.0
1995
Q1	53.8	92.4	63.7
Q2	-19.5	8.1	5.5
Q3	6.1	-13.4	-7.0
Q4	17.9	23.7	20.0
1996
Q1	-8.5	-10.4	-7.3
Q2	7.1	4.5	3.8
Q3	28.0	32.6	31.1
Q4	5.9	7.3	6.3

^1/See Appendix II for definitions. As indicated above, a measure exceeding 14 percent is considered expansionary; below 8 Percent, contractionary; and in between, neutral.

First, a regime shift occurred between the period through the first quarter of 1995 and thereafter.²¹ Following a strongly expansionary monetary stance, the CBA imposed a drastic tightening of monetary policy from mid-1995, with two of the three measures showing contractionary effects by the second quarter. From that time onwards, monetary policy was contractionary, albeit with two significant let-ups in the final quarter of 1995 and the third quarter of 1996. Thus, the second conclusion is that the stance of monetary policy was practically never neutral during the period analyzed. In fact, monetary policy displayed a stop-go cycle, with tight stance followed in a seemingly regular fashion by a loosening. Finally, as is evident from Figure 3, the monetary impulse measures are correlated with inflation. Although base money expansion in part reflects the rise in prices, this is interpreted as evidence that changes in the monetary stance drove changes in the price level during the stabilization period.

D. Dollarization²²

This section deals with dollarization—the substitution of foreign money for domestic money—and its evolution during the transition phase in Armenia. Dollarization encompasses currency substitution, the replacement of domestic currency in its means of payments function, and asset substitution, i.e. the replacement of domestic currency in its store of value, and unit of account functions. As stabilization took hold, the extent of both these declined. However, there is a serious measurement problem, since at least three important components of foreign currency holdings of residents should be taken into account: cash foreign exchange holdings, offshore foreign currency deposits of residents, and foreign exchange deposits in the banking system. It should be kept in mind that since only the third component is directly observed, the following analysis does not cover all aspects of the evolution of dollarization.

As documented in several studies on the subject, dollarization is most prevalent in economies with high levels of inflation and expectations of exchange rate depreciation as residents try to protect the real value of their wealth by holding a more stable foreign currency. As the severity of the macroeconomic imbalances diminish, it is expected that people switch back to the domestic currency. However, evidence on this second hypothesis is more mixed, as explained below. For example, in some Latin American economies, most notably Bolivia, the economy has continued to be dollarized even after the economy has enjoyed stability for several years.

The ratio of foreign currency deposits in broad money declined to around a fifth by late 1995 and stayed at this level through end-1996. While this level is substantial, it is not excessive in a small, open economy at Armenia’s stage of development. It is similar to the 15 percent level experienced in Georgia, and compares favorably with Bolivia and Vietnam, where over 80 percent of broad money is held as foreign currency deposits in spite of a stable economy. In the Armenian context, the level of this ratio reflects the risk averse behavior of individuals in a situation characterized by regional tensions and a trade blockade; the stability of the ratio may actually mask a continued decline in dollarization concurrent with remonetization due to an improving perception by the public of the health of the banking system, part of which is occurring in foreign exchange. Finally, some repatriation of deposits previously held off-shore may have also occurred, in part boosting foreign exchange deposits.

The approximate constancy of the foreign currency ratio during the second part of the stabilization period at non-negligible levels has two important implications. First, it gives rise to “network effects”: the public gets used to foreign currency substituting for the domestic currency in all its functions. The process of erosion and partial replacement of the domestic currency-based system is hard to reverse even after the economy achieves stability. Although no survey data exists, there is anecdotal evidence to support this view for Armenia: dollars can be used for transactions quite easily, especially in the rapidly expanding informal economy. A second implication is that with persistent dollarization, the CBA loses a significant portion of the noninflationary component of seigniorage, that associated with the growth of the economy. Also, if inflation is viewed as a tax on domestic money, to the extent that dollarization is present, it limits the ability of the government to raise that tax.

In addition to the negative implications mentioned so far, dollarization may also carry benefits, especially in countries like Armenia during the period analyzed, which are emerging from high inflation and unstable macroeconomic conditions. Allowing foreign currency deposits under these circumstances may speed up the process of reintermediation since agents may be more willing to return their assets to domestic banks if they can eliminate exchange rate risk by keeping them in foreign exchange. Moreover, foreign currency deposits can help achieve the desired distribution of risks and returns for domestic agents, which is particularly important for a small, relatively open economy which is at an initial stage of developing its financial markets.²³

Following Ramirez-Rojas (1985) and Rojas-Suarez (1992) we estimate the following simple econometric model to determine the factors explaining dollarization in Armenia:

where FC_t denotes the ratio of foreign currency deposits to broad money (used as a measure of dollarization); ‘trend’ represents a deterministic time trend, FC_t-1denotes the stock adjustment variable; E_pt denotes the expected change in the domestic currency in period ‘t’ vis-à-vis the dollar and u_t is a white noise error term.

In order to avoid simultaneous equation bias, the change in the exchange rate in period ‘t-1’ is used to proxy the expected change in the exchange rate in period ‘t’, and instrumental variables estimation is used to estimate the model. The dollarization and exchange rate variables were tested for non-stationarity, and the null hypothesis of a unit root was rejected for both series (see Appendix III). Since both the ratio of foreign currency to domestic broad money and the rate of change of the exchange rate are stationary processes, traditional critical values can be used to test the significance of the parameters.

Using monthly observations from January 1994 to December 1996, the following estimates for the parameters were obtained:

The figures in parentheses are the t-statistics, and the ‘*’ indicates that the coefficient is significant at the 95 percent level of confidence. Consistent with our expectations (and also with studies for other countries), the results show a positive relation between dollarization and exchange rate expectations. The coefficient of the variable representing exchange rate expectations is both positive and significant. The value of the coefficient at 0.34 suggests that if the expected rate of depreciation falls by one percent, people decrease their foreign currency holdings by one-third of one percent. The stock adjustment variable, which describes the speed at which people adjust their foreign currency holdings in each period, is also significant. The value of the coefficient indicates that people adjust their holdings of foreign currency with considerable inertia. Finally, the trend variable is not significant.

Although the relatively large size of the stock adjustment variable indicates that people adjust their holdings of foreign currency slowly, the significance of the exchange rate variable suggests that sound policies achieving sustained stability of the exchange rate can contribute to lowering the degree of dollarization. On the other hand, exchange rate volatility is likely to be interpreted as a negative signal, making it harder to reduce dollarization in the economy.