1. A brief history of Icelandic growth ^2/

Due to the richness of its resources, Iceland has enjoyed strong growth rates (averaging 3.5 percent between 1966 and 1994) and high living standards throughout the past decades (Chart 14). Currently, however, the economy is recovering from a prolonged downturn that began in 1988, caused by the over-fishing of Icelandic fish stock and a series of severe terms of trade shocks. Although yielding high export revenues, the domination of the business cycle by the fisheries has created structural problems for other sectors. As a result, diversification efforts have met limited success, and the investment to GDP ratio has been declining over the last two decades. Moreover, total factor productivity growth has slowed down markedly during the same period.

View Full Size

ICELAND OUTPUT INDICATORS

(Annual percentage change)

Citation: IMF Staff Country Reports 1995, 047; 10.5089/9781451819182.002.A001

Sources: Data provided by Icelandic authorities; and staff calculations.

• Download Figure
• Download figure as PowerPoint slide

The high dependency on fish is an inherent feature of the Icelandic economy. More than 15 percent of GDP is directly or indirectly related to the fishing business, and recessions in the fisheries have always translated into slumps in the overall growth rate, such as in 1983 or 1988. Despite a number of diversification efforts, Iceland has maintained a relatively constant 75-80 percent share of marine products in merchandise exports. ^1/ However, the high profitability in the fishing sector and the common property nature of the fish stock led to the buildup of over-capacity, and consequently to over-fishing. It takes about 4-5 years for cod to grow to catchable sizes and therefore, a significant decrease in the fish stock--as experienced after 1987--was bound to have a more than temporary effect on catch volumes.

Due to its dominating size, fluctuations in the fishing sector have usually translated into strong movements of the real effective exchange rate (REER; see Chart 14). Strong export growth was often not countered by corrective fiscal policies and led to an over-heating of the economy, promoting inflation and increases in the REER. ^2/ Appreciations of the REER were regularly countered by nominal devaluations and restrictive domestic wage policies. Over time, however, this process led to simultaneous increases of wage demands and prices, with inflation rates reaching more than 40 percent on average during the 1970s and 1980s. Only after the adoption of a fixed exchange rate regime, and in the wake of rising unemployment, could wage increases be contained to allow a sustained reduction of inflation rates.

Concerning export activities of other sectors, the high revenues of the fisheries were a mixed blessing. Iceland suffered from a Dutch disease effect, namely that fishing revenues were reflected in a high real exchange rate that hurt other export sectors, notably manufacturing. Strong fluctuations of the exchange rate acted as an additional obstacle for investments in export-oriented industries, which were already at a disadvantage of selling to a small domestic economy and producing in a location far from major export markets. Reflecting the rich thermal energy resources of Iceland, the only major export contribution of the manufacturing industry consisted in power-intensive aluminum and ferro-silicon products. More recently, however, significant improvements in competitiveness have fostered greater export diversification.

During the past two decades, Iceland has experienced a drastic fall in the gross investment to GDP ratio. From a peak of some 35 percent in 1974, investment declined to less than half this level in 1994. Although this fall can be partly attributed to the withdrawal of the Government as a major investor (often having financed ill-fated diversification projects in the 1970s), other factors have also contributed. For most of the time, investment was hindered by a volatile exchange rate and high inflation rates. Moreover, investment also suffered from the financial sector not being able to provide financing at adequate interest rates. At first, the supply of credit was limited throughout the 1970s, due to the prevailing negative real interest rates. Later on, however, the constraint fell on the credit demand side when real long-term interest rates swiftly climbed to rather high levels (of more than 10 percent by 1988) after they became positive in 1984. In addition, the past reluctance of Iceland to allow foreign investment and, in particular, foreign ownership exacerbated the situation, e.g. by inhibiting the enlargement of the power-intensive sector during the 1980s.

As a consequence of reduced investment in the manufacturing and service sectors, but also due to the buildup of over-capacity in the fisheries, the growth of productivity declined during the past decades. Total factor productivity rose on average by 1.1 percent in 1984-92, compared to 2.5 percent in 1969-1975 and 1.5 percent in 1976-83 (see Chart 14). ^1/ Due to the strong dependency of Iceland on the fishing sector, total factor productivity is closely related to the size of the fish catch. If the stock is larger, productivity increases because the average catch per vessel increases. Therefore, the slowdown in productivity growth after 1987 owes partly to the decline in the stock.

The current recovery poses some interesting questions. Notably, it is a recovery that cannot be based on a fast return to previously reached fish production levels, given the depleted cod stock. Moreover, in contrast to earlier upswings, domestic prices and wages have been stable, and unemployment remains relatively high by past standards. Therefore, it is worth investigating the sources of past and present growth in order to assess whether the current recovery can be sustained in the medium term.

2. The analytical framework

In what follows, a simple decomposition of GDP is used to derive a growth model for the Icelandic economy. The purpose of this model is to quantify the impact of changes in important variables on GDP growth. It will describe how the relative weight of factors like fish production, REER, or terms of trade can be determined and how the model has been estimated.

It is well known in the literature that the econometric analysis of factors behind economic growth poses considerable difficulties. ^2/ In particular, explanations are sought for why the production factors (labor and capital) fail to account for a large part of economic growth. Standard growth accounting models (e.g., Fischer (1993) or Ram (1985)) focus on the influence of technology shocks, human capital accumulation, inflation or budget effects on investment and productivity. They concentrate mainly on the production side and try to identify growth factors over relatively long periods. These approaches are not necessarily well suited for modelling short-term fluctuations in demand variables, such as those caused by exchange rate or real wage movements, and for the analysis of economies like Iceland’s that are strongly dependent on exports. Several papers have covered both theoretical and empirical approaches to Icelandic growth (Einarsson and Magnusson (1987), Einarsson (1992), and Herbertsson (1994)); however, they also do not contain models that could be used to answer the questions of this section. ^1/

The following model therefore pursues a slightly different approach. ^2/ In order to evaluate the impact of changes in the economic environment, the model regresses the demand components of GDP on a number of variables which are relevant to the Icelandic economy, uses the results to derive short-term elasticities, and calculates contributions to growth. The approach is based on the observed constancy of the shares of GDP demand components. Splitting GDP into domestic absorption (A), exports (X), and imports (I), the basic accounting identity becomes Y = A + X - I. The growth of the demand components can then be estimated by a simultaneous equations model.

Two econometric versions of the model have been estimated. The first version is a purely static formulation that contains variables that have been assumed to affect growth a priori. The second version has been given a dynamic structure by including a number of lagged variables. Facing the restriction that consistent time-series are only available from 1965, the set of variables was reduced by eliminating insignificant variables after a sequence of statistical tests. Consequently, the dynamic model is supported by the standard model selection criteria and is used for further analysis.

The quantitative impact of shocks can be evaluated by obtaining growth elasticities from the reduced form and by conducting an impulse response analysis. Although the approach implies that some of the endogeneity in the variables might not have been completely captured, it has yielded some valuable insights into the nature of Icelandic growth and into the mechanisms of how shocks are passed through the economy. Moreover, it has proved fairly robust with respect to different model specifications.

3. Contributions to growth

The estimation results of both model versions confirm that, besides the long-run impact of the production factors labor and capital, fish production and demand in other industrial countries are the most important factors for Icelandic growth. The quantitative effect of these and other important variables is discussed in the following. It is noted, however, that the results have to be interpreted carefully in light of the simplicity of the employed model.

Beginning with the estimation of the national accounts identity, the following relationship for the growth rates of GDP, domestic absorption, exports and imports has been estimated (standard errors in parentheses): ^1/

The weights implied by this equation have remained fairly stable over time, as could be confirmed by a set of tests on structural breaks. Moreover, the result does not violate on a statistically significant level the constraint that the true weights must add up to one. Therefore, the relationship is used with confidence for the further analysis.

Proceeding with the estimation results contained in Tables 1 and 2, the results indicate that there is a clear causal link from a number of exogenous variables to export performance, and further to domestic absorption. In fact, the final export equations basically contain a set of exogenous variables that can be grouped into three categories. First, lagged export and fish production are exogenous instruments that were solely added to capture the observed fluctuations in exports. Second, fish and aluminum production are regarded as price-inelastic variables. ^2/ Third, demand effects are captured by GDP growth of industrial countries (exogenous to Iceland), by the REER and real wages (which has been dropped in the dynamic model). The impact of the REER and tourist arrivals, the only remaining endogenous variables, is relatively small; other possibly endogenous variables were tested and have been found insignificant. ^3/

Table 1.

Estimation Results for the Static Model

Note: ^*, ^**, ^*** indicate significance on the 10, 5, and 1 percent level, respectively. N is the number of observations, and AR(1) gives the test statistic for autocorrelated residuals.

^1/Variable expressed as GDP ratio (tax, deficit) or in actual levels (interest rates).

^2/Elasticity is computed only for variables denominated in growth rates. For the remaining variables, the column contains the respective partial derivative of GDP growth. For example, an increase in the real long-term interest rate of one percentage point would decrease GDP growth by around 0.4 percentage points.

Table 1.

Estimation Results for the Static Model

	Domestic Absorption	Exports	Imports	Static Elasticity 2/
Domestic Absorption			1.4429 (1.86)*	--
Exports	0.2048 (0.89)		0.0276 (0.11)	--
Industrial Countries’ GDP		0.3529 (0.71)		0.1480
Labor Force	-0.3688 (-0.42)	0.9054 (1.18)	-0.4054 (-0.65)	0.3213
Capital Stock	1.2998 (1.98)*	-0.4625 (-0.79)	-0.3585 (-0.93)	0.6028
Fish Production		0.4942 (3.90)***		0.2073
Aluminum Production		0.0641 (1.91)*	0.0694 (1.50)	0.0039
Tourist Arrivals		0.0856 (0.89)		0.0359
Real Wage	0.2675 (1.48)	0.0424 (0.25)	-0.0002 (-0.01)	0.1575
Tax Ratio 1/	0.0009 (0.76)			0.0498
Central Government Deficit 1/			0.0028 (0.53)	-0.0918
Real Long-term Interest Rate 1/	-0.0073 (-1.05)			-0.3833
Real Short-term Interest Rate 1/	0.0001 (0.09)			0.0046
REER	0.2358 (1.33)	-0.2909 (-1.60)	0.0214 (0.11)	0.0081
Terms of Trade		0.1263 (1.02)	0.0101 (0.06)	0.0496
N	27	27	27
AR(1)	16.44**	32.17**	8.77*
Correlation Actual-Fitted	0.770	0.757	0.765

Note: ^*, ^**, ^*** indicate significance on the 10, 5, and 1 percent level, respectively. N is the number of observations, and AR(1) gives the test statistic for autocorrelated residuals.

^1/Variable expressed as GDP ratio (tax, deficit) or in actual levels (interest rates).

^2/Elasticity is computed only for variables denominated in growth rates. For the remaining variables, the column contains the respective partial derivative of GDP growth. For example, an increase in the real long-term interest rate of one percentage point would decrease GDP growth by around 0.4 percentage points.

View Table

Table 1.

Estimation Results for the Static Model

	Domestic Absorption	Exports	Imports	Static Elasticity 2/
Domestic Absorption			1.4429 (1.86)*	--
Exports	0.2048 (0.89)		0.0276 (0.11)	--
Industrial Countries’ GDP		0.3529 (0.71)		0.1480
Labor Force	-0.3688 (-0.42)	0.9054 (1.18)	-0.4054 (-0.65)	0.3213
Capital Stock	1.2998 (1.98)*	-0.4625 (-0.79)	-0.3585 (-0.93)	0.6028
Fish Production		0.4942 (3.90)***		0.2073
Aluminum Production		0.0641 (1.91)*	0.0694 (1.50)	0.0039
Tourist Arrivals		0.0856 (0.89)		0.0359
Real Wage	0.2675 (1.48)	0.0424 (0.25)	-0.0002 (-0.01)	0.1575
Tax Ratio 1/	0.0009 (0.76)			0.0498
Central Government Deficit 1/			0.0028 (0.53)	-0.0918
Real Long-term Interest Rate 1/	-0.0073 (-1.05)			-0.3833
Real Short-term Interest Rate 1/	0.0001 (0.09)			0.0046
REER	0.2358 (1.33)	-0.2909 (-1.60)	0.0214 (0.11)	0.0081
Terms of Trade		0.1263 (1.02)	0.0101 (0.06)	0.0496
N	27	27	27
AR(1)	16.44**	32.17**	8.77*
Correlation Actual-Fitted	0.770	0.757	0.765

Note: ^*, ^**, ^*** indicate significance on the 10, 5, and 1 percent level, respectively. N is the number of observations, and AR(1) gives the test statistic for autocorrelated residuals.

^1/Variable expressed as GDP ratio (tax, deficit) or in actual levels (interest rates).

^2/Elasticity is computed only for variables denominated in growth rates. For the remaining variables, the column contains the respective partial derivative of GDP growth. For example, an increase in the real long-term interest rate of one percentage point would decrease GDP growth by around 0.4 percentage points.

Table 2.

Estimation Results for the Dynamic Model

Note: ^*, ^**, ^*** indicate significance on the 10, 5, and 1 percent level, respectively. N is the number of observations, and AR(1) gives the test statistic for autocorrelated residuals.

^1/Variable expressed in actual levels.

^2/Elasticity is computed only for variables denominated in growth rates. For the remaining variables, the column contains the respective partial derivative of GDP growth. For example, an increase in the real long-term interest rate of one percentage point would decrease GDP growth by around 0.14 percentage points.

Table 2.

Estimation Results for the Dynamic Model

	Domestic Absorption	Exports	Imports	Long-run Elasticity 2/
Domestic Absorption			1.5080 (10.90)***	--
Domestic Absorption (-1)	-0.0675 (-0.67)			--
Domestic Absorption (-2)	-0.1834 (-2.50)**			--
Exports (-1)	0.4643 (4.58)***	-0.4696 (-3.29)***		--
Industrial Countries’ GDP		1.0845 (2.22)**		0.3744
Labor Force	0.2079 (0.49)	0.7682 (1.34)	-0.2535 (-0.51)	0.4322
Capital Stock	1.1225 (2.62)**	-0.4265 (-0.84)	-0.5499 (-1.63)	0.4834
Fish Production		0.5046 (4.94)***		0.2146
Fish Production (-1)		0.1169 (1.33)		--
Aluminum Production		0.0957 (3.38)***	0.0739 (2.29)**	0.0086
Tourist Arrivals		0.0411 (0.43)		0.0142
Real Wage	0.2616 (3.83)**			0.1046
Long-term Real Interest Rate 1/	-0.0033 (-2.45)**			-0.1339
REER		-0.1278 (-1.83)*		-0.0441
N	26	26	26
AR(1)	16.86**	33.77**	5.92*
Correlation Actual-Fitted	0.913	0.869	0.896

Note: ^*, ^**, ^*** indicate significance on the 10, 5, and 1 percent level, respectively. N is the number of observations, and AR(1) gives the test statistic for autocorrelated residuals.

^1/Variable expressed in actual levels.

^2/Elasticity is computed only for variables denominated in growth rates. For the remaining variables, the column contains the respective partial derivative of GDP growth. For example, an increase in the real long-term interest rate of one percentage point would decrease GDP growth by around 0.14 percentage points.

View Table

Table 2.

Estimation Results for the Dynamic Model

	Domestic Absorption	Exports	Imports	Long-run Elasticity 2/
Domestic Absorption			1.5080 (10.90)***	--
Domestic Absorption (-1)	-0.0675 (-0.67)			--
Domestic Absorption (-2)	-0.1834 (-2.50)**			--
Exports (-1)	0.4643 (4.58)***	-0.4696 (-3.29)***		--
Industrial Countries’ GDP		1.0845 (2.22)**		0.3744
Labor Force	0.2079 (0.49)	0.7682 (1.34)	-0.2535 (-0.51)	0.4322
Capital Stock	1.1225 (2.62)**	-0.4265 (-0.84)	-0.5499 (-1.63)	0.4834
Fish Production		0.5046 (4.94)***		0.2146
Fish Production (-1)		0.1169 (1.33)		--
Aluminum Production		0.0957 (3.38)***	0.0739 (2.29)**	0.0086
Tourist Arrivals		0.0411 (0.43)		0.0142
Real Wage	0.2616 (3.83)**			0.1046
Long-term Real Interest Rate 1/	-0.0033 (-2.45)**			-0.1339
REER		-0.1278 (-1.83)*		-0.0441
N	26	26	26
AR(1)	16.86**	33.77**	5.92*
Correlation Actual-Fitted	0.913	0.869	0.896

Note: ^*, ^**, ^*** indicate significance on the 10, 5, and 1 percent level, respectively. N is the number of observations, and AR(1) gives the test statistic for autocorrelated residuals.

^1/Variable expressed in actual levels.

^2/Elasticity is computed only for variables denominated in growth rates. For the remaining variables, the column contains the respective partial derivative of GDP growth. For example, an increase in the real long-term interest rate of one percentage point would decrease GDP growth by around 0.14 percentage points.

Furthermore, the equations for domestic absorption also demonstrate the importance of the export sector. First, there is a direct effect of export revenues on domestic demand in the following year (as indicated by the lagged export coefficient); and second, domestic absorption also depends on real wages which in turn have closely followed the fish production cycle in the past. However, domestic absorption is also characterized by a strong business cycle effect, leading to negative coefficients for the two absorption lags. Among other variables, only the long-term interest rate appears to have a significant impact on demand, yielding around 0.14 percent of GDP growth for every reduction of 100 basis points. This finding is broadly consistent with the observed importance of the housing sector which is particularly sensitive to changes in the long-term interest rate.

The relationship between imports and domestic absorption is clearly complementary. Even after the estimation of several different specifications of the import equation, the only remaining result is that imports are more or less completely driven by domestic demand. The exception is minor, namely that aluminum production requires a high amount of foreign raw materials, and that therefore imports increase with rising aluminum exports. Although it seems striking that imports are not dependent on the REER or the terms of trade, there are two explanations for this. First, Iceland has a small production base and is relatively strongly dependent on foreign goods, making imports less price-sensitive. It is rather the overall demand that is affected by changes in external prices, and this is reflected in the estimates. Second, terms of trade have mostly moved in line with the REER (see Chart 14), which in turn was mostly dependent on the fortunes of the fisheries. Therefore, import fluctuations are also caused indirectly by changes on the export side. In the model, this effect is also contained in the coefficient of domestic absorption.

Finally, it has been confirmed that capital stock and labor force have the expected impact on growth. However, because both variables change little in the short-run, they are generally used to explain the long-term trends of growth. The results are therefore consistent with the fact that the decline of average growth rates over the past decades has occurred simultaneously with a drop in the investment ratio and a fall in the female participation rate. The estimated capital elasticity in the dynamic model is smaller than in the static model, indicating that shocks to the capital stock are smoothed over the business cycle. ^1/ Even in the dynamic model, however, the growth elasticities of both variables are large, being estimated at 43 percent for the labor force, and 48 percent for the capital stock. These values are in line with the results of Herbertsson (1994) who found output shares of 42 and 46 percent, respectively, for the period 1946-80.

The impulse response of growth on a one percent shock in domestic absorption is negative, reaching -0.1 percent after four years. This reflects mainly an increase in the demand for imports which is not matched by a growing export volume. The impulse response of growth on a shock in exports is positive but small, yielding an accumulated additional growth of 0.04 percent after four years.

Summing up, the most important growth factors besides labor and capital are export-related. The Icelandic economy is not only strongly sensitive to developments in the fishing sector but also, primarily in the long-run, to demand conditions in other industrial countries. Compared to the fisheries, other export sectors like aluminum production and tourist arrivals have a relatively minor impact (as expressed by the long-run elasticity of growth of both sectors which is a mere tenth of the one for fish production). On the domestic side, real wages and real long-term interest rates have a relatively strong impact (with a long-run elasticity and derivative, respectively, of some 10 percent). On the other hand, a number of variables did not turn out to be important for growth, in particular fiscal variables. It has to be noted, however, that the results are derived from an approach that is based solely on econometric evidence, mixing supply and demand variables and thus remaining valid only for a limited time period. The results should therefore be interpreted carefully. However, the outcome has been fairly robust over several model specifications, and therefore gives a valuable insight into the impact magnitude of several variables on Icelandic growth.

4. The current recovery

Both an analysis of the ongoing recovery and a growth projection underline the continuing dependency of Iceland on the fishing sector. Growth of fish production was the strongest factor causing the turnaround of real GDP growth after 1992. Moreover, a growth forecast based on the estimated relationship yields a sustainable growth rate of 2.4 percent by the end of the decade.

Comparing the predicted values from a dynamic simulation with actual growth rates (Chart 15), it is evident that exports have replaced domestic absorption as the main driving force behind GDP growth since 1992. Domestic absorption and imports were much stronger during 1990-91 than simulated by the model whereas, at the same time, exports performed relatively poorly. According to the simulation, growth rates were kept from becoming negative mainly by the unpredicted high growth rates of the domestic economy. However, both domestic absorption and imports fell below their simulated values in 1992, whereas exports approached their forecasted level. In 1994, exports were higher than simulated, whereas imports increased in line with predictions and both GDP and domestic absorption remained sluggish.

View Full Size

ICELAND DOMESTIC ABSORPTION, EXPORTS AND IMPORTS: SIMULATED VS. ACTUAL

(Annual percentage change)

Citation: IMF Staff Country Reports 1995, 047; 10.5089/9781451819182.002.A001

Sources: Data provided by Icelandic authorities; and staff calculations.

• Download Figure
• Download figure as PowerPoint slide

The deviations between actual and simulated values can be explained by two straightforward arguments, owing to recent developments which were not captured by the model. First, the strong fluctuation of domestic absorption around its predicted path is mostly caused by an exceptional growth rate in 1991. The increase in domestic absorption in 1991 was caused by an over-proportional growth of private consumption, following a strong nominal wage increase, and accompanying growth in public consumption and gross investment (the latter due to aircraft purchases). Moreover, the reduction in inflation rates promoted increased consumer confidence. Subsequently, higher unemployment rates diminished the effective labor force and contributed to a decline in private consumption. ^1/

Second, the exceptional export performance in 1994 is also due to a relatively new development. Besides increasing fish production, other sectors have also started to contribute to export growth, notably the manufacturing and service sectors. Fishery equipment and services have picked up strongly, and the service sector, especially, has evolved only recently. Consequently, exports have so far seen a rare period of three subsequent years of recovery.

The relative impact of the explaining variables on the ongoing recovery is contained in the tabulation below. The numbers indicate the share of each variable in the improvement of GDP growth between 1992 and 1994, which amounted to an aggregate 5.4 percentage points. The shares have been computed by applying the long-term elasticities to the difference between the 1994 and 1992 growth rates of the explaining variables. Altogether, the variables account for 66.2 percent of the growth increase (the dynamics of the model capture another 35.1 percent). The overall forecast error is thus 1.3 percent (or 0.07 growth percentage points), hinting at a relatively well specified model. The upswing owes most to the fishing sector which contributed some 20 percent, whereas labor force and capital stock increases are relatively less important than earlier periods (resulting from lower participation rates of females and falling investment). The worldwide recovery, improvements in the REER, and tourist arrivals have also contributed to the recovery, albeit in a less prominent role.

The model-based growth projection for the next five years (Chart 16) shows that, conditional on the continuation of sound economic policies, Iceland should be able to sustain a growth rate of close to 2.5 percent. This result is mainly based on the assumption of an optimal fishing policy, allowing a slow increase of fish production from the year 1997 (reaching a growth rate of 3 percent by 1999), and a stabilizing growth rate of the capital stock, reaching 2 percent in 1997 and 2.5 percent by 1999. Apart from that, domestic policies are assumed to facilitate real wage increases along with productivity growth (1.5 percent) and the long-term interest rate is modelled to increase slightly. The GDP of industrial countries is expected to increase annually by 2.5 percent, fostering growth of aluminum exports and tourism. The projection does not assume that a new aluminum smelter is being built during the next 4 years. The REER has been kept unchanged over the forecast period.

View Full Size

ICELAND MODEL PREDICTIONS FOR DOMESTIC ABSORPTION, EXPORTS, IMPORTS AND GDP ^1/

(Annual percentage change)

Citation: IMF Staff Country Reports 1995, 047; 10.5089/9781451819182.002.A001

Source: Data provided by Icelandic authorities; and staff calculations.^1/ Shaded area indicates standard prediction error.

• Download Figure
• Download figure as PowerPoint slide

Share of Variables in the Current Recovery (1992-94)

Share of Variables in the Current Recovery (1992-94)

		Growth percentage	Percent
Industrial Countries’ GDP		0.44	8.3
Labor Force		0.76	14.1
Capital Stock		0.83	15.3
Fish Production		1.14	21.1
Aluminum Production		0.04	0.7
Tourist Arrivals		0.19	3.6
Real Wage		-0.07	-1.3
Long-term Real Interest Rate		-0.01	-0.1
REER		0.24	4.4
	Total share of explaining variables	3.57	66.2
Dynamic effects		1.90	35.1
Residual		-0.07	-1.3
	Total growth	5.40	100.0

View Table

Share of Variables in the Current Recovery (1992-94)

		Growth percentage	Percent
Industrial Countries’ GDP		0.44	8.3
Labor Force		0.76	14.1
Capital Stock		0.83	15.3
Fish Production		1.14	21.1
Aluminum Production		0.04	0.7
Tourist Arrivals		0.19	3.6
Real Wage		-0.07	-1.3
Long-term Real Interest Rate		-0.01	-0.1
REER		0.24	4.4
	Total share of explaining variables	3.57	66.2
Dynamic effects		1.90	35.1
Residual		-0.07	-1.3
	Total growth	5.40	100.0

Export growth is projected to slow in 1995, reflecting both the still depleted fish stocks and the current uncertainty about further catches in international waters. After 1997, with the fishing sector picking up, export growth strengthens and reaches 2.8 percent by the end of the projection period. The export outlook might be better if the non-traditional export sector continues to perform well but, judging from the impulse response analysis, the impact on growth would be small. Domestic absorption is projected to grow by some 3.5 percent in 1995, based on stronger investment, but will decline slightly in 1996, provided wage increases remain moderate and fiscal policy is geared towards further consolidation. Thereafter, it is projected to grow steadily with rising exports. Imports move basically in line with domestic absorption. As a result, GDP growth would rise to about 2.5 percent by 1999.

Annex: The Modified Growth-accounting Model

The model originates from the accounting equation

with Y = GDP, A = domestic absorption, X = exports, and I = imports, all in real terms. Expressed in growth rates, it holds approximately that

where g denotes the growth rate as first difference in logs. The estimation of (2) yields a vector of estimated shares, denoted γ.

To take account of the external shocks (e.g., terms of trade and fish catch), a system of equations is formulated that comprises each of the single, sectoral growth rates. Feedback between the different sectors is allowed for by including the respective growth rates in the equations. The model in its static form then looks like:

where g = (g_A g_X g_I)’. The x-vector contains a set of explaining variables which are assumed to have an impact on overall growth through one or more of the three sectoral equations. The model has been set up to be fully identified, and can be estimated by full information maximum-likelihood. In order to calculate the elasticities, the estimated parameters are inserted into the Γ and Π matrices, and considering that GDP growth is modeled as g_Y = γ’g, one obtains

with the elasticities calculated as -γ’ Γ^-1 Π.

In a second step, lags are added to the system to yield the dynamic version, specified as:

Long-run elasticities are calculated by setting g = g_-1 = g_-2 and x = x_-1. Consequently, the elasticities result as -γ’ (Γ + Γ₁ + Γ₂)^-1 (Π + Π₁).

In general, the estimations have performed relatively well. As Icelandic data were only available on an annual basis, the remaining degrees of freedom were quite small, preventing the use of more sophisticated techniques. Neither could the problem of autoregressive errors be resolved satisfactorily, nor the high correlation between residuals across equations. Dynamic simulations have been conducted by using actual values for the explaining variables and simulating only the components of output growth.

1. A Primary gap approach

In this section we make use of the primary gap indicator to assess the sustainability of fiscal policy in Iceland. The primary gap indicates the amount of fiscal consolidation that would be necessary in a given year in order to stabilize the debt-GDP ratio at its prevailing level.

Following the approach discussed in Blanchard (1990), which starts from the government’s budget constraint ^1/, it can be shown that, expressed in terms of ratios to GDP, the change in the net debt to GDP ratio satisfies the following equation:

where g is the real GDP growth rate, π is the inflation rate and lower case letters denote ratios to GDP of variables defined in the footnote below. Assuming some constant values for r and g, equation 2 implies that for fiscal policy to be sustainable at the current debt ratio, the present value of primary surpluses b discounted by the difference between the real interest rate and the real growth rate (r-g), must be equal to the initial level of debt (d).

This implies that if the Government is to meet its intertemporal budget constraint, primary budget deficits now will need to be offset by primary surpluses later, if the Government is not to resort to ever increasing levels of borrowing to finance its overall operations. ^2/

denote the constant primary surplus that would stabilize the debt-to-GDP ratio. The difference between this and the actual primary surplus, b, measures the deviation of current policies from a sustainable fiscal path. This indicator is commonly referred to as the primary gap and will be denoted by:

Starting from equation (3), and given the official fiscal projections for 1995 and 1998, we consider alternative scenarios based on different assumptions regarding real GDP growth, real interest rate and domestic inflation, and derive the magnitude of fiscal consolidation that would be necessary in order to stabilize the debt-to-GDP ratio at its 1994 level. Assuming real interest rates to be maintained at 5 percent, the primary gap is derived for the cases when the economy will grow at 1.5, 2, 2.5, and 3 percent, respectively. Those four cases are also examined assuming that the real interest rate will be at 4 percent. In all cases, the inflation rate is assumed to be maintained at 2.5 percent.

The results of this exercise, displayed in the tabulation below, indicate that in all cases, fiscal adjustment would be required to stabilize the net debt at 30 percent of GDP in 1995. Moreover, it is shown that, based on prudent assumptions regarding economic growth and interest rate development, the official objective of budgetary balance by 1998 is consistent with fiscal sustainability.

Primary gap

(In percent of GDP)

Source: Staff calculations.

Primary gap

(In percent of GDP)

	r = 5 percent		r = 4 percent
	1995	1998	1995	1998
case 1: g = 1.5%	1.2	0.0	1.0	-0.2
case 2: g = 2%	1.0	-0.2	0.8	-0.4
case 3: g = 2.5%	0.8	-0.4	0.6	-0.6
case 4: g = 3%	0.7	-0.5	0.5	-0.7

Source: Staff calculations.

View Table

Primary gap

(In percent of GDP)

	r = 5 percent		r = 4 percent
	1995	1998	1995	1998
case 1: g = 1.5%	1.2	0.0	1.0	-0.2
case 2: g = 2%	1.0	-0.2	0.8	-0.4
case 3: g = 2.5%	0.8	-0.4	0.6	-0.6
case 4: g = 3%	0.7	-0.5	0.5	-0.7

Source: Staff calculations.

2. A saving-investment approach

A continuing financial deficit of the public sector is sustainable over time only if it contributes to investment and growth in the economy as a whole. In Iceland, the treasury deficit should therefore be examined against the overall development of saving and investment. As shown in Chart 17, the levels of domestic investment and saving have tended to decline over the past decade. Also, until 1993, total investment exceeded national saving by a relatively wide margin, implying a large current account deficit. Another aspect worth noting in the saving-investment analysis is the relative role of the Treasury in contributing to the national saving-investment gap. Chart 18 shows that since the mid-1980s, the treasury saving gap has averaged about 2 percent of GDP. The rest of the economy (private sector and municipalities together) has also invested on average more than it has saved, the gap, however, being in recent years somewhat smaller than that of the Treasury.

View Full Size

ICELAND SAVING AND INVESTMENT

(In percent of GDP)

Citation: IMF Staff Country Reports 1995, 047; 10.5089/9781451819182.002.A001

Source: National Economic Institute.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

ICELAND TRESURY SAVING GAP

(In percent of GDP)

Citation: IMF Staff Country Reports 1995, 047; 10.5089/9781451819182.002.A001

Source: Ministry of Finance, Treasury Finances.

• Download Figure
• Download figure as PowerPoint slide

While the relationship between fiscal policies and national saving and investment developments are complicated and not well understood, it would appear safe to conclude that better balanced treasury finances could contribute favorably to the overall growth prospects in Iceland by making more room for private investment. The accompanying table below summarizes the results of some exercises about the requirements on public finances stemming from a saving and investment analysis of the economy.

The analysis is based on the following assumptions. The private sector gross saving rate is assumed to be the same as on average for 1990-94 (16.5 percent). Assuming an increase in real gross investment by up to 6 percent a year and a real GDP growth of about 2 percent a year, this would imply that the total investment level will increase from 16.5 percent of GDP in 1995 to 18.2 percent by 1998. The increase is assumed to be about the same in relative terms in the private sector. An investment level of at least 18 percent of GDP might be an appropriate target based on the experience of industrial countries over the last two decades. The external current account surplus is assumed to remain at its 1995 projected level of 1 percent of GDP.

The result of this simple exercise is that public sector financial saving would have to be improved by about 0.7 percent of GDP in 1996 and a further 1,2 percent and 0.7 percent in 1997 and 1998 respectively. The public sector would have to reach zero saving gap in 1997 and contribute to financial saving with 0.7 percent of GDP in 1998, in order for the economy as a whole to generate enough saving to finance the targeted level of investment without a deficit on the external current account.

Saving and Investment

(In percent of GDP)

Source: Staff calculations.

Saving and Investment

(In percent of GDP)

		Outcome	Projected	Target
		1994	1995	1996	1997	1998
Gross saving		17.7	17.3	17.8	18.6	19.2
	Private	16.5	16.5	16.5	16.5	16.5
	Public	1.2	0.8	1.3	2.1	2.7
Gross Investment		15.6	16.5	16.8	17.6	18.2
	Private	12.7	13.8	14.3	15.5	16.2
	Public	2.9	2.7	2.5	2.1	2.0
S/I balance		2.1	0.8	1.0	1.0	1.0
	Private	3.8	2.7	2.2	1.0	0.3
	Public	-1.7	-1.9	-1.2	0.0	0.7

Source: Staff calculations.

View Table

Saving and Investment

(In percent of GDP)

		Outcome	Projected	Target
		1994	1995	1996	1997	1998
Gross saving		17.7	17.3	17.8	18.6	19.2
	Private	16.5	16.5	16.5	16.5	16.5
	Public	1.2	0.8	1.3	2.1	2.7
Gross Investment		15.6	16.5	16.8	17.6	18.2
	Private	12.7	13.8	14.3	15.5	16.2
	Public	2.9	2.7	2.5	2.1	2.0
S/I balance		2.1	0.8	1.0	1.0	1.0
	Private	3.8	2.7	2.2	1.0	0.3
	Public	-1.7	-1.9	-1.2	0.0	0.7

Source: Staff calculations.

3. Official projections of the medium-term fiscal outlook

The Ministry of Finance’s assessment of Iceland’s medium-term fiscal prospects was incorporated in the October 1995 budget proposal. Two scenarios were presented: a baseline which assumed unchanged policies and showed a rising deficit over the medium term, and an alternative scenario based on measures that were projected to bring the budget into balance by 1998.

In the baseline scenario, the development of revenues and expenditures were calculated based on current laws and regulations taking into account general demographic trends, but in the absence of any specific measures on behalf of the Government. Given the increased longevity of the Icelandic population, on unchanged policies, the rapid increase of transfer payments--amounting currently to 40 percent of total expenditures--would continue unabated; operating expenditures would also continue to claim a significant proportion of government spending. This would lead the treasury deficit to rise to 5 percent of GDP by 1998, which coupled with a low nominal GDP growth assumed over this period, would push the debt ratio close to 60 percent. Moreover, in view of recent deficit-widening measures associated with subsequent fiscal concessions made in connection with the February 1995 wage agreement, the deficit would (by staff estimates) under this baseline scenario reach 6 percent of GDP, implying a debt ratio of over 60 percent.

The budget document noted that the consequences of a rising deficit for the overall macroeconomic picture of the economy would be deleterious. Interest rates would increase, hampering economic growth, the stability of prices and exchange rates could be threatened, the competitive position of industry would be eroded, the current account would turn into a deficit and foreign debt would build up again. The document sets out an alternative scenario designed to avert these pitfalls. Under this alternative (target) scenario, specific measures would be implemented to eliminate the deficit by 1998. To this end, the budget proposal articulated broad plans for expenditure restraint, including a 5 percent cumulative reduction in public consumption through increased accountability and use of contracting out. Second, transfer payments were to fall by almost 6 percent by targeting the largest categories of transfers (social security and agricultural sector). This would be achieved by reevaluation of priorities and increased efficiency to better match the need of the lowest income groups. Finally, public investment would be targeted to decline. On these assumptions, expenditures would fall by 7 percent in real terms, while revenues would remain constant, bringing a budgetary balance by 1998. The treasury’s gross debt would fall from 49 percent of GDP in 1994 to 46 percent in 1998 (Chart 19).

View Full Size

ICELAND CENTRAL GOVERNMENT GROSS DEBT

(In percent of GDP)

Citation: IMF Staff Country Reports 1995, 047; 10.5089/9781451819182.002.A001

Source: Ministry of Finance, Treasury Finances.

• Download Figure
• Download figure as PowerPoint slide

Although the goal of budgetary balance and the expenditure measures to achieve it in the target scenario are indicative plans to inform discussion of medium-term budgetary issues, they clearly show the positive effects of credible measures of balancing the budget. In the target scenario, economic growth would be higher, climbing to 2 percent, interest rates are expected to be much lower, which lead to higher private investment and domestic demand increases without disturbing the stability of prices and exchange rates. Finally, the balance on the current account would be positive, further lowering the foreign debt.

Recent reforms

Although the new system represented a clear improvement over the pre-1987 tax system with respect to both revenue-raising and resource allocation, substantial distortions remained and the tax structure was not well suited to increased competition with Europe. The corporate tax burden was higher than the average elsewhere in Europe ^1/, while the personal tax rates with numerous allowances and credits were lower than the European average of 15 percent. ^2/ This led to further reforms in the tax system during 1992-94. The main tax change in 1992 was the abolition of the import equalization tax. ^1/ Additionally, further adjustments were made on the personal income tax system, aimed at broadening the tax base. Tax exemptions for seamen were tightened and changes were made in the child benefit system in favor of higher benefits to lower income parents.

In 1993, the Government made major changes in the corporate tax structure in order to reduce business costs and boost competitiveness. Corporate income tax rates were lowered first from 45 to 39 percent and then to 33 percent in 1994, but at the same time the tax base was broadened by the abolition of corporate investment tax credits (the so-called investment funds). The turnover tax, an important source of revenue for the municipalities, was also abolished. In the context of the wage negotiations in spring of 1993, several tax concessions were also made, among which was the lowering of the VAT on food products from the higher rate of 24.5 percent to 14 percent as of January 1, 1994. Personal income taxes were raised by 0.4 percentage points and the employer’s social security tax by a similar amount. Finally, the system of tariffs and excises was subject to extensive changes in the context of the single market provisions the EEA agreement.

Major changes in 1994 included the introduction of a 14 percent VAT on hotel services, a reduction of the corporate income tax rate from 39 to 33 percent, the reduction of interest rebates for residential housing and a tax reshuffling between the central Government and the municipalities. ^2/ Apart from the plan to introduce by 1996 a capital income tax in the form of nominal withholding tax on interest income, the 1995 budget contains no new discretionary tax changes.

Comparative analysis

Compared to other European countries, the burden of taxation in Iceland is quite low. Tabulation A below shows that at about 36 percent of GDP, tax revenue in Iceland is much lower than the European average of 45.5 percent. The recent lowering of the VAT-on foods from 24.5 percent to 14 percent and the effects of numerous exemptions and rebates in the income tax system have contributed to making Iceland one of the most lightly taxed countries in Western Europe. Unlike other industrial countries, a significant share of general government’s tax revenue is attributed to indirect taxes. As shown in Tabulation B, in 1994 indirect taxes accounted for over 55 percent of general government total tax revenue in Iceland. This compares with 41 percent and 45 percent respectively in Sweden and Norway. It appears however that the relative importance of this form of taxation has been diminishing, at least for the central Government, reflecting recent changes in the tax system.

Tabulation A. General Government Current Receipts

(As a percentage of GDP)

Source: OECD.

Tabulation A. General Government Current Receipts

(As a percentage of GDP)

	1990	1991	1992	1993	1994
Sweden	63.3	60.2	59.8	58.3	56.2
Norway	56.3	55.2	55.0	54.4	55.2
Finland	50.7	52.5	53.3	53.7	54.0
Iceland	36.0	37.3	37.8	36.5	35.9
Average Europe	43.9	44.6	45.1	45.5	45.5

Source: OECD.

View Table

Tabulation A. General Government Current Receipts

(As a percentage of GDP)

	1990	1991	1992	1993	1994
Sweden	63.3	60.2	59.8	58.3	56.2
Norway	56.3	55.2	55.0	54.4	55.2
Finland	50.7	52.5	53.3	53.7	54.0
Iceland	36.0	37.3	37.8	36.5	35.9
Average Europe	43.9	44.6	45.1	45.5	45.5

Source: OECD.

Tabulation B. Indirect Taxes

(As a percentage of general government total taxes)

Source: OECD.

Tabulation B. Indirect Taxes

(As a percentage of general government total taxes)

	1990	1991	1992	1993	1994
Sweden	42.1	47.2	44.7	42.3	41.9
Norway	39.7	39.9	42.3	44.0	45.1
Finland	46.2	46.6	47.7	49.5	45.0
Iceland	64.8	61.2	60.3	56.8	56.0

Source: OECD.

View Table

Tabulation B. Indirect Taxes

(As a percentage of general government total taxes)

	1990	1991	1992	1993	1994
Sweden	42.1	47.2	44.7	42.3	41.9
Norway	39.7	39.9	42.3	44.0	45.1
Finland	46.2	46.6	47.7	49.5	45.0
Iceland	64.8	61.2	60.3	56.8	56.0

Source: OECD.

Measured in terms of government spending per GDP, Iceland has a relatively small government sector. General government expenditures, as a ratio of GDP, amounted to roughly 40 percent compared with the European average of 51 percent. However, the functional distribution of central Government is quite typical and in line with other industrial countries. The largest category, accounting for about 40 percent of total expenditures in recent years covers welfare, including social security, pensions, health and public housing.

General Government Total Outlays

(As a percentage of GDP)

Source: OECD.

General Government Total Outlays

(As a percentage of GDP)

	1990	1991	1992	1993	1994
Sweden	59.1	61.3	67.2	71.8	67.4
Norway	53.8	55.4	57.4	57.1	56.5
Finland	45.4	53.9	59.1	60.8	58.6
Iceland	40.1	41.0	41.4	40.6	40.4
Average Europe	47.6	49.0	50.3	52.0	51.5

Source: OECD.

View Table

General Government Total Outlays

(As a percentage of GDP)

	1990	1991	1992	1993	1994
Sweden	59.1	61.3	67.2	71.8	67.4
Norway	53.8	55.4	57.4	57.1	56.5
Finland	45.4	53.9	59.1	60.8	58.6
Iceland	40.1	41.0	41.4	40.6	40.4
Average Europe	47.6	49.0	50.3	52.0	51.5

Source: OECD.

1. Introduction

Indexation has been a feature of the Icelandic financial system for many years, and its scope expanded sharply in the late 1970s. Rapid inflation led to large scale financial disintermediation, with total deposits in the deposit money banks, measured as a ratio of GDP, falling below 20 percent in 1978 from an average of 40 percent in the 1960s. The Central Bank had made an attempt in the mid-1970s to link interest rates informally to inflation so as to bring real rates to positive levels. Although interest rates moved in the right direction, it took a general indexation scheme stipulated in laws in 1979, to bring about a positive real interest rate. Following the passage of the Economic Management Act in 1979, almost all long-term securities, some loans, deposits and debt instruments were linked to a price index (the Credit Terms Index) so that changes in the index instantaneously alter the amount of financial obligations in question.

Interest rates, which had been inflexible in Iceland for most of the 1970s remained so even after the institutionalization of indexation. Initially, real interest rates as well as nominal rates on non-indexed instruments continued to be administratively set by the Central Bank. With the financial liberalization process that began in 1984, market forces were given an increasing role in interest rate determination. The Central Bank Act of 1986 left commercial banks free to set practically all their own interest rates. Following the deregulation of interest rates, deposit and lending rates rose, as banks increasingly competed among themselves and with other institutions for financial savings. Real interest rates rose sharply both on indexed claims and non-indexed claims such as non-indexed bank loans.

While financial indexation appears to have contributed to the lowering of inflation from chronically high levels, and therefore helped bring about positive real interest rates and increased savings, there is an ongoing debate in Iceland as to whether the mechanism of indexation should be dismantled altogether or simply reformed, e.g by changing the composition of the Credit Term Index to which financial instruments are indexed. A prominent view among policymakers in Iceland is that markets should be left to decide when and how to dismantle financial indexation. More generally, indexation of financial assets at the short-end are viewed as having more drawbacks than merits. There is also a belief that high interest rates in Iceland might to some extent be attributed to the indexation of financial assets and liabilities.

This appendix examines the role of financial indexation in affecting the behavior of interest rates in Iceland. Arguments in favor of and against indexation of financial assets are brought forth to show that the debate regarding whether indexation leads to an increase in interest rates or not, boils down to the question of whether the financial markets are efficient under rational expectations. Starting from this view, empirical tests are undertaken using data on indexed and non-indexed bank rates to examine the extent to which the indexation of debt instruments has affected the level of interest rates in Iceland. The main findings are that financial indexation has not induced an increase in interest rates and that the interest rate spread between indexed and non-indexed assets is an indicator of future inflation.

2. Institutional background

Financial indexation has a long history in Iceland. The practice of indexation in Iceland started in 1955 with a 25 percent indexation of general housebuilding loans and a full indexation of the Compulsory Savings of young people in 1957. In 1961, the Fisheries Loan Fund started granting loans that were partly linked to the exchange rate while the Treasury began issuing indexed saving bonds linked to the Building Cost Index in late 1964. Although a law was passed in 1966 concerning the indexation of financial assets, indexation was introduced very slowly and did not cover to any extent financial assets outside the public credit system until 1972.

Indexation of credit granted by the Pension Funds to the Investment Credit Funds began in 1972. A year later, some of the Investment Credit Funds introduced indexation of credit granted out of funds obtained from the pension funds. In 1977, the Central Bank permitted the pension funds to adopt a partial indexation of 40 percent on their general loans, provided that the interest rate be no higher than 11.5 percent and the amortization period at least 10 to 15 years. The same year, it was decided by law that the pension funds should use at least 40 percent of their disposable funds to buy fully indexed long-term bonds from the Investment Credit Funds. These were the main fields in which indexation was applied until the Government announced a system of general price indexation of saving and credit in April 1979. ^1/

Subsequently, 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. The minimum loan maturity was initially set at four years, shortened to two and half years in 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.

Until the late 1980s, Iceland had followed a policy of low and inflexible nominal interest rates. With persistent inflation during much of the early period, real interest rates were often negative. The real interest rate on total deposits at the DMBs was -3.8 percent in the period 1960-72, while the average real interest rate on bank lending was -1.3 percent. During 1973-77, the inflation rate accelerated and the real interest rates on deposits and lending fell to -17.7 percent and -16.8 percent, respectively. With accelerating inflation and the alarming fall in the ratio of deposits to GDP in DMBs, a more flexible interest rate policy was called for at the beginning of the 1980s. As a result, banks introduced “switching term deposits”--whose terms switched between ordinary and indexed rates according to which was more favorable to the depositor--in order to maintain competitiveness.

The development of real interest rates on indexed obligations from the beginning of the liberalization process can be seen in Chart 20. It shows the interest rates on non-indexed bank loans and government bonds from 1987-94. The turnaround in real rates on indexed obligations and loans was even more dramatic as real rates went from being highly negative to being highly positive in the course of a few years. (See Chart 20) The change from negative to positive interest rates had became imperative as financial saving was being heavily eroded and disintermediation had become a serious problem. The spread of indexation, high interest rates and deregulation resulted in a strong growth of the capital market in the mid-1980s.

View Full Size

ICELAND REAL INTEREST RATES

(In percent)

Citation: IMF Staff Country Reports 1995, 047; 10.5089/9781451819182.002.A001

Sources: Central Bank of Iceland, Economic Statistics, Annual Report; and data provided by Icelandic authorities.

• Download Figure
• Download figure as PowerPoint slide

3. Empirical evidence

Whether and to what extent indexation has affected the level of interest rates in Iceland remains an unresolved question. ^1/ As a prelude to the empirical work that follows, we first examine the nature of the relationship between the returns on indexed and non-indexed bonds in a market in which transactors are free to choose between these two instruments. If the redemption value and coupon payments of indexed bonds are to be adjusted in accordance with the development of the price level, then the rate of interest carried by indexed and non-indexed bonds cannot be identical, but must reflect these differences.

In equilibrium, the rates of interest established must be such that the expected yield to be obtained from purchasing either type of bond is the same for all holders, given their expectations about the rate of price increase, and subject perhaps to an allowance for any change in their risk position which transactors may think is involved if indexed rather than non-indexed bonds are held. Thus, the relationship between the two rates of interest is approximated by the Fisher equation namely that r = i-π^e where π^e is the expected rate of inflation. For a given ex ante real rate of interest, the Fisher effect dictates that a higher nominal interest rate applies to financial transactions if prices are expected to rise.

The case in favor of financial indexation can be presented by considering an environment in which errors in inflationary expectations are not made systematically and markets operate efficiently. If the price expectations of all bond holders are uniform and the authorities offer indexed bonds in replacement of maturing non-indexed bonds on terms which exactly reflect these price expectations, no upward or downward pressure on the interest rates should be experienced. In this case, if one disregards the risk involved in the possible non-fulfillment of price expectations, it will be a matter of indifference to any bondholder whether he/she holds indexed or non-indexed bonds. ^2/ Under these conditions, the Fisher effect implies that the evolution of interest rates in an indexed system cannot deviate systematically from the evolution of interest rates in a non-indexed system.

The case against financial indexation is often based on the hypothesis that full Fisherian transmission never takes place because expectations underestimate inflation systematically, or that there is no efficient arbitrage in financial markets to eliminate the inflation risk component of interest rates. Suppose the authorities issue indexed bonds on terms which, for that volume, are in line with market expectations, then because of the lack of efficient arbitrage, the lenders who expect the most rapid price inflation will not be able to remove themselves from the indexed market so as to bring about an interest rate parity across markets. Under these conditions, indexation will force nominal interest rates to be higher than under normal market conditions.

As these arguments show, the conflicting views about financial indexation are founded on different notions regarding the efficiency of credit markets and the manner in which individuals formulate expectations about future inflation. In economies where financial markets are well organized and competitive, and where market participants formulate expectations rationally, indexation cannot result in systematically higher interest rates than a system without indexation, the intuition being that indexation eliminates the inflation-risk component of interest rates.

The case of Iceland

This section presents empirical tests to assess the effectiveness of indexation in eliminating the inflation-risk factor of nominal interest rates in Iceland. The claim that indexation does not validate higher interest rates than a non-indexed system in Iceland is examined by undertaking two tests: one that explores the degree of market efficiency and another that extracts the information conveyed by the differential between indexed and non-indexed interest rates regarding future increases in inflation. Both tests are conducted jointly with the hypothesis that agents formulate their expectations rationally, i.e., that their projections of future inflation are not systematically biased.

Consider a financial market of the Icelandic-type, where indexed and non-indexed instruments co-exist. A lender has the option of holding an indexed instrument with real yield r or buy a non-indexed instrument with a nominal rate i. The nominal return on the indexed bond is linked to the lender’s expectation of the Credit Terms Index (CTI) at time t. As discussed in the preceding section, if the market is efficient, the following will emerge as an arbitrage condition: ^1/

$\begin{array}{cc}\mathrm{(}\mathrm{1}\mathrm{+}\mathrm{N}\mathrm{O}\mathrm{M}\mathrm{)}\mathrm{=}\mathrm{(}\mathrm{1}\mathrm{+}\mathrm{R}\mathrm{E}\mathrm{A}\mathrm{L}\mathrm{)}\mathrm{(}\mathrm{1}\mathrm{+}{\mathrm{E}}_{\mathrm{t}}\mathrm{T}\mathrm{)}& \mathrm{\left(}\mathrm{1}\mathrm{\right)}\end{array}$

where NOM is the monthly nominal interest rate on 90-day non-indexed bank loans, REAL is the real rate on indexed bank loans, T = ΔCTI/CTI is the change in the Credit Term Index ex-post, and E_t denotes the expectation operator conditional on information available at date t. It can be shown (see Annex) that under conditions of risk aversion, the above arbitrage condition may not hold at equality due to the presence of a risk premium. Non-indexed bonds may need to be offered at a premium because the two assets are not perfectly substitutable or because of risk aversion on the part of the investors. In the case of Iceland, the risk premium could be due to the uncertainty associated with the possibility of changes in the CTI. To test the empirical content of this theory, we estimate the following equation:

$\begin{array}{cc}\mathrm{N}\mathrm{O}{\mathrm{M}}_{\mathrm{t}}\mathrm{=}\mathrm{R}\mathrm{P}\mathrm{+}\mathit{\beta }\mathrm{(}\mathrm{R}\mathrm{E}\mathrm{A}{\mathrm{L}}_{\mathrm{t}}\mathrm{+}{\mathrm{T}}_{\mathrm{t}}\mathrm{)}\mathrm{+}{\mathit{\mu }}_{\mathrm{t}}& \mathrm{\left(}\mathrm{2}\mathrm{\right)}\end{array}$

where RP is a time-invariant risk premium. The test will analyze jointly efficient arbitrage (unbiasedness hypothesis), a time-invariant risk premium, and a particular hypothesis of how expectations are formed. It is assumed here that expectations by agents in the Icelandic financial markets are formed rationally, i.e., on the basis of all publicly available information. If expectations are rational, forecasting errors are random variables that follow a white-noise process.

To be consistent with the hypothesis mentioned, the coefficient estimates should be RP^* ≠ 0, β^* = 1, if there is efficient arbitrage under rational expectations and the error term μ_t must be a serially uncorrelated random variable. The null hypothesis can be written as

${\text{H}}_0\text{\hspace{0.5em}:}\text{RP}\ne \text{\hspace{0.5em}0}\text{\hspace{0.5em},}\beta \text{\hspace{0.5em}=}\text{\hspace{0.5em}1}\text{,}\left\{{\mu }_{\text{t}}\right\}\text{\hspace{0.5em}serially}\text{\hspace{0.5em}uncorrelated.}$

The estimation of equation (2) is done using monthly data on the effective real rates paid on indexed bank loans and non-indexed secured loans from 1986 to 1995. The estimation results over different subperiods are summarized in the tabulation below. The results for the period as a whole, support the hypothesis that there is efficient arbitrage in Icelandic financial markets in the sense that β^* is not statistically different from 1. This implies that, corrected for the liquidity premium, changes in the effective return of indexed loans are reflected in almost equally proportional changes in the effective return of non-indexed loans. The hypothesis of a time invariant risk premium (RP ≠ 0) also could not be rejected at the 5 percent level, suggesting that there is a risk premium attached to the rates on non-indexed loans. The regression as a whole has a high explanatory power and a low standard error, and it produces residuals that are serially uncorrelated.

Estimates of Regression Equation

Note; Numbers in brackets are t-statistics for the null hypothesis that the corresponding coefficient is not significantly different from zero while t (β=1) are t-statistics for the null hypothesis that the coefficient β is not different from 1.

Estimates of Regression Equation

Period	RP	β	SE	R2	DW	t (β=1)	F
1988-1994	3.5 (3.21)	0.82 (9.35)	4.06	0.70	1.55	2.01	202.24
1988-1993	3.2 (1.71)	0.83 (9.17)	4.75	0.66	1.49	1.97	137.61
1988-1992	2.35 (1.33)	0.85 (9.45)	4.65	0.65	1.51	1.85	109.39
1988-1991	1.92 (0.70)	0.86 (7.53)	5.20	0.55	1.49	1.22	56.81
1988-1990	0.64 (0.16)	0.91 (5.82)	5.93	0.49	1.42	0.57	33.89

Note; Numbers in brackets are t-statistics for the null hypothesis that the corresponding coefficient is not significantly different from zero while t (β=1) are t-statistics for the null hypothesis that the coefficient β is not different from 1.

View Table

Estimates of Regression Equation

Period	RP	β	SE	R2	DW	t (β=1)	F
1988-1994	3.5 (3.21)	0.82 (9.35)	4.06	0.70	1.55	2.01	202.24
1988-1993	3.2 (1.71)	0.83 (9.17)	4.75	0.66	1.49	1.97	137.61
1988-1992	2.35 (1.33)	0.85 (9.45)	4.65	0.65	1.51	1.85	109.39
1988-1991	1.92 (0.70)	0.86 (7.53)	5.20	0.55	1.49	1.22	56.81
1988-1990	0.64 (0.16)	0.91 (5.82)	5.93	0.49	1.42	0.57	33.89

Note; Numbers in brackets are t-statistics for the null hypothesis that the corresponding coefficient is not significantly different from zero while t (β=1) are t-statistics for the null hypothesis that the coefficient β is not different from 1.

Chart 20 depicts the evolution of the real interest rates on indexed and non-indexed bank loans. The real interest rate on non-indexed bank loans is measured by the average nominal interest rate during each quarter, deflated by price increases over the quarter. This is therefore an ex-post measure of the real interest rate and fluctuates considerably due to changes in the rate of inflation and rigidities in nominal interest rates. The chart shows that both rates follow similar trends with the rates on indexed loans being systematically lower than those on non-indexed loans in recent years. The higher real rates on indexed loans compared to non-indexed loans in the period 1980-88 reflect the premium that was associated with the agents’ perception of the risk involved in the variance of monthly inflation during that period.

A related question is to what extent does the interest rate differential between indexed and non-indexed bonds reflect inflationary expectations in Iceland. The informational content of the spread between indexed and non-indexed interest rates can be extracted by estimating a functional relationship that follows from the Fisher effect. ^1/ If the differential reflects inflation expectations, then assuming that expectations are fully realized, a regression of actual inflation on the spread should produce a coefficient estimate not statistically different from one. Such an equation was estimated by ordinary least squares using quarterly data from 1988 to 1995. In this case, the regression output is summarized as follows:

$\begin{array}{cccc}\mathrm{I}\mathrm{N}\mathrm{F}\mathrm{=}& \mathrm{-}\mathrm{0.231}\mathrm{+}& \mathrm{1.11}\mathrm{(}\mathrm{N}\mathrm{O}\mathrm{M}\mathrm{-}\mathrm{R}\mathrm{E}\mathrm{A}\mathrm{L}\mathrm{)}\mathrm{+}& {\mathit{\mu }}_{\mathrm{t}}\\ & \mathrm{\left(}\begin{array}{c}\mathrm{0.29}\end{array}\mathrm{\right)}& \mathrm{\left(}\mathrm{14.85}\mathrm{\right)}\hfill & \end{array}$

$\mathrm{A}\mathrm{d}\mathrm{j}\mathrm{.}{\mathrm{R}}^{\mathrm{2}}\mathrm{=}\mathrm{0.89}\mathrm{D}\mathrm{W}\mathrm{=}\mathrm{1.81}\mathrm{S}\mathrm{E}\mathrm{=}\mathrm{2.94}\mathrm{F}\mathrm{=}\mathrm{220.60}$

The results show that the hypothesis that the coefficient estimate is not different from 1 cannot be rejected at the 5 percent significance level. Furthermore we are unable to reject the null hypothesis that the constant term is not significantly different from zero, suggesting that changes in the differential between indexed and non-indexed rates provide an indicator of future inflation. Chart 21 depicts the actual and predicted values of inflation produced by the model during the period considered. The chart illustrates the predictive power of the regression based on the spread between indexed and non-indexed interest rates to forecast the evolution of the uncovered increase in inflation.

View Full Size

ICELAND ACTUAL AND PREDICTED INFLATION RATE

(Quarterly change at an annual rate, in percent)

Citation: IMF Staff Country Reports 1995, 047; 10.5089/9781451819182.002.A001

Source: Central Bank of Iceland; and staff calculations.

• Download Figure
• Download figure as PowerPoint slide

4. Current views of indexation in Iceland

One of the arguments given for the introduction of indexed bonds in Iceland was that this kind of security would stimulate private voluntary saving. It is reckoned that full indexation, by enabling savers to know how much their savings will be worth in the future in terms of purchasing power, has had a positive effect on domestic saving in Iceland. Total deposits and bonds issued by banks have recovered to over 40 percent of GDP and pension funds’ assets and other financial savings have been inflation-protected. As shown in Chart 22, indexed deposits rose from next to nothing in 1980 to over 60 percent of total deposits in 1994. The introduction of indexed bonds was also proposed because of its expected positive effects on fighting inflationary pressures. While the lowering of inflation in Iceland may be attributed to a number of factors, including a successful exchange rate policy, there is little doubt that indexation of financial assets played a major role in reducing the disruptive impact of past high inflation on the financial system.

View Full Size

ICELAND INDEXED DEPOSITS AND CREDITS

(In percent of total)

Citation: IMF Staff Country Reports 1995, 047; 10.5089/9781451819182.002.A001

Source: Central Bank of Iceland; and data provided by Icelandic authorities.

• Download Figure
• Download figure as PowerPoint slide

These merits notwithstanding, there have been calls in recent years to reduce the scope of indexation in the financial market. One problem associated with indexation is its adverse effects on interest rate formation in the banking system. Interest rate decisions by banks have been complicated by the mismatch between their indexed assets and indexed liabilities. ^1/ Banks’ indexed liabilities (mainly deposits) have been far bigger than their indexed assets, leading them to raise nominal interest rates higher than they would have been, because of the risk premium it entails. ^1/ Another problem is that, because indexation is associated with a high-inflation mentality, it carries allocational costs. On this ground, and in view of the low level of inflation rate, there is a case to be made for reducing indexation in the Icelandic financial system.

Attempts to move in this direction started in July 1988, when the Government banned direct indexation of the principal value of financial instruments with maturities shorter than two years and 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 saving account in each six-month period. In October 1993, as part of its effort to reduce interest rates, the Government announced a package of reforms which included the reduction of indexation on short-term instruments; the minimum maturity for indexed loans was reduced from 3 to 2 years, and deposits could only be indexed for a maturity of over 1 year.

How far the dismantling of the indexation should go depends on the progress towards disinflation. Given the crucial role played by the mechanism of indexation in creating a stable financial system, a gradual removal of it through market forces might be the appropriate approach. ^2/ A sudden removal of indexation could be interpreted as an attempt by the Government to resort to inflationary finances, and therefore could jeopardize the hard-won gain in inflation performance of recent years. A prominent view in Iceland is that indexation at the short-end has had more drawbacks than merits as it tended to be market distorting, whereas at the long-end indexation has proved to be more beneficial. ^3/ Therefore, one proposal is the early introduction of non-indexed government securities with longer maturities than the present two years, as a first step toward the progressive withering of indexation provisions in the Icelandic financial system.

5. Conclusions

The purpose of this appendix has been to look into the mechanism of financial indexation in Iceland and its influence on the operation of the financial system, particularly with regard to interest rate formation. Despite calls for dismantling financial indexation, the claim that Iceland’s stubbornly high real interest rates of recent years is to some extent attributable to indexation, could not be supported by the evidence presented in this paper. Therefore, the root cause of high real interest rates in Iceland may be found elsewhere, for instance in the fiscal position, saving performance, and a relatively large external debt, A prolonged period of restricted financial activities and capital controls in the past and inadequate competition in the banking system may also have affected the level of real interest rates in Iceland.

Annex

Consider a dynamic stochastic economy which has a single good each period whose quantity varies stochastically over time and across consumers. Each period t, for t = 0, 1, …, T the economy experiences an event z_t. Let c_t(z^t) denote the consumption of the single good at date t given history z_t and by P(z^t) its price, so that P_t+1 (z^t+1)/P_t(z^t) = π(z^t+1) represents the inflation rate. Each period, y(z^t) units of the consumption good become available per capita.

There is a government in this economy who consumes g_t(z^t) units of the consumption goods, collects lump-sum taxes T(z^t) in period t if history z^t occurs and issues indexed bonds B_t(z^t) with a real yield r(z^t) known at time t. At time t, each agent in the economy has access to a one-period securities market where indexed government bonds B(z^t) are traded alongside with privately issued non-indexed bonds L(z^t) which pay off a nominal return i_t(z^t), but whose real return v(z^t+1) = i(z^t) - π(z^t+1) is uncertain at time t.

The representative consumer in this economy chooses his consumption of the single good (cⁱ), asset positions (bⁱ, lⁱ) so as to maximize the expected utility function given by