Norway: Selected Issues

This Selected Issues paper analyzes inflation in Norway with a view to shedding light on this surprising development and the possible near-term course of inflation, using statistical and econometric analyses. The paper reviews recent developments of monetary policy and inflation in Norway, applies statistical and econometric tools to identify factors influencing inflation, and describes the implications of the analysis for policymaking. Using data for six advanced small open economies explicitly targeting inflation, the paper examines empirically whether deviations of the exchange rate from their equilibrium levels systematically affect the conduct of monetary policy.

Abstract

This Selected Issues paper analyzes inflation in Norway with a view to shedding light on this surprising development and the possible near-term course of inflation, using statistical and econometric analyses. The paper reviews recent developments of monetary policy and inflation in Norway, applies statistical and econometric tools to identify factors influencing inflation, and describes the implications of the analysis for policymaking. Using data for six advanced small open economies explicitly targeting inflation, the paper examines empirically whether deviations of the exchange rate from their equilibrium levels systematically affect the conduct of monetary policy.

III. The Norwegian Government Petroleum Fund and the Dutch Disease29

“The current cost level in the Norwegian business sector is adapted to an expansion of the petroleum sector and a steady phasing-in of petroleum revenues into the mainland economy. Costs rose sharply from the mid-1960s to the mid-1970s and reached a very high level. In subsequent years, costs have varied around this level. After a period, we will be able to cover a smaller share of our imports using current petroleum revenues and drawings on the Petroleum Fund. Competitiveness must then have to be improved. It may have to be brought back to around the level prevailing at the end of the 1960s prior to Norway’s emergence as an oil nation.”

——Svein Gjedrem, Norges Bank Governor

1. Norway’s strategy regarding the transformation of its large petroleum wealth into other assets will be key to its long-term prospects. Norwegian policymakers face a set of interrelated issues: insulating the budget from changes in petroleum prices and extraction rates; intergenerational equity in the use of petroleum wealth; the fiscal pressures of population aging, which are expected to mount significantly in the coming decades; and the potential crowding out effects that rapid spending of oil wealth might bring. This last issue—the so-called Dutch disease problem—is examined in this chapter.

2. In general, the Dutch disease can manifest itself via three channels: the factor-moving effect, the spillover-loss effect, and the spending effect (Box III-1). This chapter focuses on the last of these—the spending effect—because fiscal policy can potentially mitigate it by saving petroleum revenue rather than spending it, whereas the others are structural in nature (Davis and others, 2003). In the spending effect, the use of the income from natural resource sales boosts domestic demand and crowds out traded-goods industries, notably manufacturing. The result may be lower immediate welfare, but also a permanent reduction in the industrial base, even after the resources have all been extracted, because of the loss of human capital, for example.

3. The chapter is organized as follows. Section I describes Norway’s policy response to rising petroleum revenues; Section II briefly reviews the indicators of the Dutch disease in Norway; and Section III analyzes empirically whether Norway might suffer from the Dutch disease and, by implication, whether the fiscal policy response has helped. In summary, Section II finds mixed evidence of Dutch disease effects. Section III finds some empirical evidence of these effects, and consistent with a policy of mitigating the Dutch disease, that fiscal policy does not respond to oil booms.

Effects of Resource Wealth: Factor Moving, Spending, and Spillover-Loss Effect

Corden (1984) used a three-sector (the booming sector, the tradable sector, and the non-traded sector) small economy model to explain the factor moving and spending effects of resource wealth:

Factor-Moving Effect. On the supply-side, labor shifts to the booming sector from the other sectors after an exogenous increase in the value of the booming sector’s (oil-producing sector) output raises the marginal product of labor in that sector. This shift causes a contraction of the tradable sector. In addition, this factor movement leads to an increase in the price of non-traded sector because, given other conditions, the supply of non-tradable goods declines. Since the prices of tradables are exogenously determined in the world markets, the rise in the prices of nontradables is equivalent to a real appreciation, which puts further pressure on the tradable sector. This effect generally depends on the elasticity of substitution in the production technologies in each sector and the smoothness of labor shifts across different sectors.

Spending Effect. On the demand side, the resource boom increases income in the country and stimulates demand for all goods via the income effect. Since the price of tradables is set in the world markets, this additional demand raises the relative price of non-tradables to tradables. This is equivalent to a real appreciation, which again leads to labor shifts from the tradable to the nontradable sector. As a result, the output level of the non-booming tradable sector contracts.

In addition, the literature has identified a spillover-loss effect. This results when the crowding-out of the non-resource tradables sector reduces total productivity, because learning by doing (know-how) benefits are higher in the tradables sector than in the non-tradable sector (Van Wijnbergen, 1984; Krugman, 1987; Sachs and Warner, 1995, Gylfason and others, 1997; and Torvik, 2001).

A. Norway’s Strategy for Managing Oil Wealth

4. Norway’s petroleum production and exports have increased rapidly. The country started oil production in the North Sea in 1971, but petroleum operations did not create cash flows to the state until 1975. Production and exports of oil and gas then picked up, and Norway is now the third largest oil exporter in the world. Although oil production has recently declined slightly, rising gas production has offset this. The 2005 national budget predicts that the production of oil and gas will peak around 2010 and decline gradually after that.

5. Oil revenues have contributed to Norway’s large fiscal and current account surpluses since the late 1970s. Driven by oil revenues, the general government budget recorded large surpluses for most of the 1980–2004 period, whereas the non-oil budget was mostly in deficit (Figure III-1).30 In the 2005 budget, revenues from petroleum activities are expected to amount to about 20 percent of the mainland GDP. Oil exports have averaged 13 percent of GDP since 1978, ensuring large current account surpluses even though the nonoil current account has been in deficit.

Figure III-1.
Figure III-1.

Norway: Petroleum Production, Exports and Revenues.

Citation: IMF Staff Country Reports 2005, 197; 10.5089/9781451829778.002.A003

6. The Norwegian authorities established the Government Petroleum Fund (GPF) in 1990. The GPF, which is formally a government account at Norges Bank, receives most of the petroleum revenue and invests it in financial assets abroad. It, therefore, can insulate the budget from changes in petroleum income and preserve the assets for use by future generations (Skancke, 2003). One such use is to finance growing old-age pensions, and this year the government proposed to link the GPF formally to old-age pensions.31 No transfers to the GPF took place until 1995 because of low net oil income and large oil-related expenditures. Since then, however, assets of the GPF have increased rapidly, as oil production and prices picked up and the government’s oil related investment declined. In the beginning of 2005, the market value of the GPF was estimated at Nkr1,016 billion or 78 percent of mainland GDP. The 2005 budget projects that the market value of the fund will reach NOK 2,103 billion or 128 percent of the mainland GDP in 2010 (Figure III-1).

7. In 2001 (effective for the 2002 budget) the policy of saving petroleum revenue for the future was formalized in the fiscal guidelines. Within these guidelines, the key rule sets the non-oil central government structural deficit to the long-run real return on the GPF, assumed to be 4 percent. The guidelines allow temporary deviations from the rule over the business cycle and in the event of extraordinary changes in the value of the GPF. The guidelines were meant to serve a number of purposes: the capital of the GPF is preserved for future generations; some petroleum revenue is currently being spent; and the budget is insulated against sharp changes in petroleum revenues.

B. Indications of Dutch Disease in Norway

8. The indicators of the Dutch disease in Norway are mixed, perhaps reflecting the success of the GPF/fiscal rule policy approach (Figure III-2). A number of indicators suggest the absence of an effect. The CPI-based real effective exchange rate (REER) of the krone has been broadly stable since the late 1970s. Norway’s non-oil exports, as percent of mainland GDP, have also been broadly stable. While the share of the manufacturing valued added in GDP has fallen substantially, this decline has been no more marked than that typically found in other developed economies over the same period. Set against these benign indicators, however, are large increases in the unit-labor-cost-based REER of the krone since 1994 and a relatively low export-to-output ratio in manufacturing, both suggesting a loss of competitiveness. Finally, the traded-goods sector faced severe difficulties when the currency appreciated substantially in both real and nominal terms in 2002, indicating the sector is sensitive to real exchange rate movements.

Figure III-2.
Figure III-2.
Figure III-2.

Selected Indicators

Citation: IMF Staff Country Reports 2005, 197; 10.5089/9781451829778.002.A003

9. Empirical results have also been mixed. Brunstad and Drystad (1997) found that labor which is regionally and occupationally close to the booming petroleum sector, experienced nominal wage increases relative to other types of labor, suggesting that oil wealth had a factor-moving effect. Regarding the spending effect, Hutchison (1994) concluded that the energy boom had no long-run influence, while Larsen (2004) suggested that Norway might have had some symptoms of Dutch disease in the late 1990s.

C. Empirical Analysis

10. Rather than measuring the Dutch disease effect directly, the empirical approach taken here considers, from two different angles, the potential for a Dutch disease effect. The first angle is to estimate cointegration vectors to capture the long-run effects of oil prices and government spending on manufacturing and manufacturing exports. High oil prices that translate into lower manufacturing activity would suggest a classic Dutch disease problem. The second is to estimate a vector autoregression model to examine whether oil prices raise government expenditures, a key channel of the spending effect of the Dutch disease.

11. Before undertaking these exercises, it is worth considering whether Norwegian households are Ricardian. If they were, then fiscal policies, including Norway’s fiscal guidelines and the GPF, would be offset by household saving behavior. By the same taken, Ricardian households would also take into account the relevant intertemporal effects of petroleum spending, weakening the justification for the GPF. Econometric analysis, provided in Appendix III-I, suggests that Norwegian households are partly but not fully Ricardian, implying that the GPF is potentially an effective and useful policy instrument.

Cointegration analysis

12. The Dutch disease arises, via the spending effect, when domestic spending, stimulated by an oil boom, causes a real appreciation and a contraction of the exposed sector. This sub-section estimates cointegration vectors to analyze the long-run effects of an oil boom (proxied by oil prices, poil) on the exposed sector (proxied by value-added of the manufacturing sector, ym). After much experimentation, other relevant variables affecting manufacturing output are found to be aggregate Norwegian economic activity (real mainland GDP, y, and the GDP deflator, Def), and foreign demand (European real GDP, yEU, and the European GDP deflator, DefEU, translated through real Norwegian manufacturing exports, mEX). The model can thus be written as follows:

A(L)zt=α+et(1)

where  zt=(ytm,yt,mtEX,Deft,gt,DeftEU,ytEU,ptoil).32

13. Quarterly data from OECD spanning the 1978–2004 period are used, in logarithms and seasonally adjusted. Unit root test shows that all variables are I (1), and the results of the co-integration test suggests that there are four co-integrating vectors, suggesting the following relationships (Table III-1).33 Higher oil prices are indeed associated with a decline in manufacturing value added (with an elasticity of 0.04), indicating the Dutch disease effect, although they are also related to higher overall mainland GDP (vector 1). Higher oil prices are also associated with lower EU GDP, which reduces Norwegian manufacturing exports (vectors 2 and 3). Finally, higher oil prices raise Norwegian inflation as well (vector 4). However, this last vector also suggests that higher oil prices are associated with lower government expenditure. This, in turn, would imply that the government sector is not the transmission mechanism for the Dutch disease.34 This result could reflect the prudent fiscal policy followed over the years by the Norwegian authorities: that is, such a policy has effectively cut off this transmission mechanism. To explore this issue further, the short-run dynamics of oil prices and government expenditure are estimated.

Table III-1.

Co-integration Vectors

article image

VAR analysis

14. To measure the effect of oil prices on government expenditures, a VAR model using the same variables as in equation (1), but in differences, is estimated assuming the following structural relationships among its error terms:

eym=α1uym+α2uy+α3umEX+α4uDef+α5ug+α6uDefEU+α7uyEU+α8upOil
ey=α9uy+α10umEX+α11uDef+α12ug+α13uDefEU+α14uyEU+α15upOilemEX=α16umEX+α17uDef+α18ug+α19uDefEU+α20uyEU+α21upOileDef=α22uDef+α23ug+α24uDefEU+α25uyEU+α26upOil(2)
eg=α27ug+α28uDefEU+α29uyEU+γupOil
eDefEU=α30uDefEU+α31uyEU+α32upOil
eyEU=α33uyEU+α34upOil
epOil=α35upOil

where u stands for actual shocks to the system (which are not contemporaneously correlated and have unit variance), which are translated to observed shocks e in equation (2). The equations can be re-written compactly:

et=[α1α2α3α4α5α6α7α80α9α10α11α12α13α14α1500α16α17α18α19α20α21000α22α23α24α25α260000α27α28α29γ00000α30α31α32000000α33α340000000α35]ut

This structure assumes that: a shock to real oil prices is exogenous; the price level and output level in foreign countries are exogenous to Norway; a shock to the mainland GDP deflator depends on government expenditure among other factors; exports of manufacturing depends on real exchange rate and foreign demand; and mainland GDP depends on all factors other than manufacturing output. Estimating the covariance matrix of e by the vector autoregressive system yields the estimator of the elements of the matrix in equation (2).

15. The most important parameter in equations (2), which measures the response of the error term of government expenditure to that of real oil prices, is not statistically significant. 35 In other words, the estimates are consistent with the view that government expenditure does not react to an energy boom. This result suggests that the objectives of insulating the budget from oil revenue fluctuations and the economy from fiscally induced Dutch disease effects have been met.

Appendix III-1 Ricardian Equivalence

1. The test of Ricardian Equivalence uses the indirect Euler equation approach. Using the model developed by Himarios (1995), aggregate consumption responses, taking account of the possibility of a finite horizon and of liquidity constraints of households, leads to the following consumption function:

Ct=α[(1+r)At1+Σj=0(1μ1+r)jEtYtj](A1)

where At-1 is the stock of real assets outstanding at the end of period t -1, r is the constant real return on these assets, µ is the constant probability of dying of each cohort, Yt is the real after-tax labor income and Et is the expectation operator conditional on information known in period t, α is the propensity of consumption out of total wealth. The aggregate budget constraint is:

At=(1+r)At1+YtCt(A2)

when a fraction (λ) of income goes to liquidity-constrained households, the consumption functions expressed by the observable variables are (Himarios 1995):36

Ct=(1+r1μ)(1α)Ct1αμ(1+r1μ)At1+λYtλ(1+r1μ)(1α)Yt1+ut(A3)
Ct=(1+r1μ)[1α(1μ)]Ct1αμ(1+r)21μAt2+λYt(1+r1μ)[λα(λμ)]Yt1+ut(A4)

The null hypothesis (Ricardian equivalence) is that households have infinite time horizon and do not face liquidity constraints: that is µ = λ = 0.

2. Annual SNA data from 1960 to 2003 are used. Private final consumption expenditure is used as a measure of consumption. (Real) after-tax disposable income, including social transfers, instead of after-tax labor incomes in other empirical studies, is used. This choice requires adjustments to equations (A-3) and (A-4) because after-tax disposable income includes capital income. All variables are deflated by the non-oil GDP deflator and normalized to per capita terms. The sum of gross government financial liabilities (including monetary base) and capital stock in the private sector, computed based on the permanent inventory method, is used as a proxy for real assets. Since data on gross government liabilities are available only after 1970, the series is iteratively extended to 1960 using net lending/borrowing data.37 In addition, dummy variables capture the introduction and changes in the VAT tax rates (1969, 1970, 2002 and 2003), as these create wedges between the marginal rate of substitution between different periods, the banking crisis during 1988-92, and earlier financial liberalization (1985–87). All variables are I(1), according to standard tests.

Empirical Literature on the Ricardian Equivalence

Since Barro’s (1974) classic paper, there has been a huge and growing body of literature on Ricardian equivalence following two types of testing strategies, the indirect test and the direct test. The result is still inconclusive (Bernheim 1987, Seater, 1993, and Elmendorf and Mankiw, 1998). The indirect test focused on testing the validity of such critical assumptions of Ricardian equivalence as the permanent income hypothesis, intergenerational redistribution of wealth (bequest motive), liquidity constraints, distributional effects, and distortionary taxation. The direct test examined the impact of fiscal variables (taxes, social security contributions and government debt) on the rest of the economy (consumption, saving, exchange rate, and interest rates).

The results of the indirect test throw doubt on the validity of critical assumptions of the Ricardian equivalence hypothesis. Most studies of the permanent income hypothesis confirm that consumption to some extent depends on current income. Many individuals are widely considered to face liquidity constraints, which limit consumption smoothing.

The results of the direct test are “ultimately inconclusive” (Elmendorf and Mankiw 1998). Bernheim (1987) concluded the hypothesis does not hold, but Seater (1993) concluded it holds “approximately.” Lucke (1999) examining Germany, concluded that there is hardly any evidence against the Ricardian proposition whereas the theoretical model that implies Ricardian equivalence is strongly rejected. Giorgioni and Holden (2001) found that for six countries (Israel, Italy, Korea, Singapore, Tanzania, and the United Kingdom) deficits would not have a positive effect upon private consumption, providing some support in favor of Ricardian equivalence. Marinheiro (2001) got ambiguous results for the Portuguese economy. Doménech, and others (2000) reported that Ricardian equivalence did not work in 18 OECD countries, since private saving compensated for only a small fraction of the budget deficit, and Drakos (2001) found the same result for Greece.

Mixed or inconclusive results of the direct test are in part caused by problems in methodology in econometric tests, as noted by Elmendorf and Mankiw (1998) and Seater (1993). These papers pointed out the following four problems in the literature of the direct test for Ricardian equivalence: the treatment of expectation, simultaneity, multi-collininarity, and limited power of the test.

  • Expectations. The behavior of forward-looking households depends on expectations of fiscal policy, not just the measures of current fiscal policy that are included in the regressions. If government actually follows the tax-smoothing hypothesis, the current level of taxation is the best prediction of the tax policies in the future. However, in this case, a significant negative coefficient on current taxes in consumption does not necessarily contradict Ricardian equivalence because higher current taxes indicates higher future taxes too.

  • Simultaneity. Some studies estimate the consumption function with ordinary least squares. However, this strategy is valid only when the shocks to consumption do not affect fiscal policy or other right hand side variables. Other studies use instrumental variables to avoid this problem, but finding persuasive instruments is generally very difficult.

  • Multi-collininarity. This makes the estimates unstable because of large standard deviations. One remedy to this problem is to use a longer estimation period, but this strategy may face the risk of structural changes of an economy.

  • Little power to distinguish between the Ricardian and conventional views of fiscal policy. As highlighted by Poterba and Summers (1987), using a life-cycle model with plausible structural parameters suggests individuals live long enough to make the assumption of infinite horizon a good approximation for analyzing short-run saving effects. Hence, it may be statistically quite difficult to measure propensities of consumption for a change in government tax policy, even though the difference between a small and zero marginal propensity of consumption is economically critical in the long run, since a short-run drop in savings can accumulate to a large long-run decline in the capital stock.

There are two possible ways to mitigate these problems: the Euler equation approach and the aggregate consumption function approach with instrumental variables. The Euler equation approach is collapsed to the test of the permanent income htypothesis, an indirect test. The instrumental variables approach is used to estimate regressions in the direct approach. Although it is generally very difficult to find good instruments, Cardia (1997) proposed productivity be used as an instrument for output. This was based on tests of ssssimulated data, where she found that distorting taxes reduced labor supply and output (an effect unrelated to the issue of Ricardian equivalence), which biased the estimate of the Ricardian effect of taxes on consumption.

18. 2SLS is used to avoid endogeneity problems caused by contemporaneous values of the right side variables. After-tax disposable income in period t is instrumented by GDP developments in EU (including lag), real oil productions (including lag), and real assets at period t − 2. The results of the first stage regression clearly show no signs of the problem of weak instruments. Exogeneity of instrumental variables is also checked by the over-identification test, which confirms that the choice of instrumental variables is appropriate.

19. Ricardian equivalence is rejected. The coefficients of after-tax disposable income in the current period are significantly positive in both estimated equations: a one krone increase in disposable income due to changes in taxes or transfers raises consumption by about 0.4 krone. Second, the joint null hypothesis of no liquidity constraints and infinite time horizon is rejected (“p-value” in Appendix I-1). However, examination of the underlying structural parameters (Appendix I-2), while implying a substantial fraction of households face liquidity constraints and confirming the rejection of the joint null, also suggest an implausibly high (though barely significant) probability of death and a low marginal propensity to consume. While these results are common,38 they may also suggest measurement error (Marinheiro, 2001).

Appendix I-1.

Results of the Model

article image
Appendix I-2.

Estimated Parameters

article image

APPENDIX III-2

Appendix II-1

Relevant Statistics from the Estimation of the VAR Model with Restrictions Imposed as in Equations 2 and 3.

Estimated Matrix:

article image
Standard Errors
article image
z-Statistic
article image

References

  • Barro, Robert J. 1974: “Are Government Bonds Net Wealth?” Journal of Political Economy, 82 (6), 10951117.

  • Bernheim, B. Douglas 1987:Ricardian Equivalence: An Evaluation of Theory and Evidence,” in NBER Macroeconomics Annual 1987.

  • Brunstad, Rolf Jens, and Jan Morten Drystad 1997, “Booming Sector and Wage Effects: An Empirical Analysis on Norwegian Data,” Oxford Economic Papers, 49 (1), 359380.

    • Search Google Scholar
    • Export Citation
  • Cardia, Emanuela 1997:Replicating Ricardian Equivalence Tests with Simulated Series,” American Economic Review, 87 (1), 6579.

  • Corden, W. M. 1984:Booming Sector and Dutch Disease Economics: Survey and Consolidation,” Oxford Economic Papers, 36 (3), 359380.

    • Search Google Scholar
    • Export Citation
  • Davis, J. M. and others ed. (2003): Fiscal Policy Formulation and Implementation in Oil-Producing Countries, (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Domenéch, Rafael, David Taguas, and Juan Varela 2000, “The Effects of Budget Deficit on National Saving in the OECD,” Economics Letters, 69, 377383.

    • Search Google Scholar
    • Export Citation
  • Drakos, Kostas 2001, “Testing the Ricardian Equivalence Theorem: Time Series Evidence from Greece,” Journal of Economic Development, 26 (1), 11.

    • Search Google Scholar
    • Export Citation
  • Elmendorf, Douglas W. and N. Gregory Mankiw 1998, “Government Debt” in Handbook of Macroeconomics.

  • Giorgioni, Gianluigi, and Ken Holden 2001, “Some Further International Evidence on Ricardian Equivalence: A VECM Approach,” Working Paper, Center for International Banking, Economics and Finance, Liverpool John Moores University, 2001.

    • Search Google Scholar
    • Export Citation
  • Gylfason, T., Herbertson, T.T. and Zoega, G. 1997, “A Mixed Blessing: Natural Resources and Economic Growth,” CEPR Discussion paper No. 1668, London.

    • Search Google Scholar
    • Export Citation
  • Himarios D. 1995, “Euler Equation Tests of Ricardian Equivalence,” Economics Letters, 48, 165171.

  • Hutchison, Michael M. 1994, “Manufacturing Sector Resiliency to Energy Booms: Empirical Evidence from Norway, the Netherlands, and the United Kingdom,” Oxford Economic Papers, 46 (2), 311329.

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, 2001, Norway: Selected Issues, IMF Country Report No. 01/34, (Washington: International Monetary Fund).

  • Krugman, P. 1987, “The Narrow Moving Band, the Dutch Disease and the Competitive Consequences of Mrs. Thatcher: Notes on Trade in the Presence of Dynamic Scale Economies,” Journal of Development Economics, Vol. 27.

    • Search Google Scholar
    • Export Citation
  • Larsen, Erling Roed 2004, Escaping the Resource Curse and the Dutch Disease? When and Why Norway Caught up with and Forged ahead of its Neighbours, Statistics Norway, Discussion Papers No. 377, May 2004.

    • Search Google Scholar
    • Export Citation
  • Lucke, B 1999, “Econometric Tests for Ricardian Equivalence: Results for Germany,” Unpublished paper.

  • Marinheiro, C.F. 2001, “Ricardian Equivalence: an Empirical Application to the Portuguese Economy,” Unpublished paper.

  • Poterba, James M., and Lawrence H. Summers 1987, “Finite Lifetimes and the Effects of Budget Deficits on National Saving,” Journal of Monetary Economics, 20, 369391.

    • Search Google Scholar
    • Export Citation
  • Romer, D. 1996, Advanced Macroeconomics, McGraw-Hill.

  • Sachs, Jeffrey D. and Andrew M. Warner 1995, “Natural Resource Abundance and Economic Growth,” NBER Working Paper 5398.

  • Seater, John J. 1993, “Ricardian Equivalence,” Journal of Economic Literature, 33 (1), 142190.

  • Skancke, Martin 2003, “Fiscal Policy and Petroleum Fund Management in Norway,” in Fiscal Policy Formulation and Implementation in Oil-Producing Countries ed. By J.M. Davis, R. Ossowski, and A. Fedelino (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Tsalik, Svetlana 2003, Caspian Oil Windfalls: Who Will Benefit? ed. Rober Ebel, Open Society Institute, Central Eurasia Project.

  • Torvik, R. 2001, “Learning by Doing and the Dutch Disease,” European Economic Review, Vol. 45.

  • Van Wijnbergen, S. 1984, “The Dutch Disease: A Disease After All?,” Economic Journal, Vol.94.

29

Prepared by Etibar Jafarov and Kenji Moriyama.

30

The state receives revenues from oil enterprises through taxes and royalties, fees, its direct financial interest in the petroleum sector (SDFI), and dividends from state shares Statoil and Norsk Hydro (see IMF 2001).

31

The GPF is widely admired and the authorities of several other oil producing countries, in designing their own oil funds, extensively studied Norway’s experience (Tsalik 2003).

32

Different order of the elements in vector z was also computed, but the main conclusions do not change.

33

Based on the Schwarz criterion, three lags of the variables were used.

34

This does not, of course, rule out the possibility that household consumption or business investment increase in reaction to the oil boom to produce a Dutch disease effect. While this mechanism is beyond the scope of this chapter, recall that Appendix III-1 found Norwegian consumers to be partly Ricardian, implying some such response.

35

The estimated coefficient for γ is -0.0002, standard deviation is 0.0017, and t-stat is -0.1057, nearly zero. Full results are reported in Appendix II.

36

See Romer (1996) for the review of literature on the liquidity-constraint hypothesis.

37

Using financial assets omitting government debt does not change the main results.

38

Himarios (1995), using US data, reported the estimates (α,μ,λ) = (0.054,0.145,0.449) for equation (3) and (α,μ,λ) = (0.056,0.159,0.419) for equation (4), where all estimated coefficients were statistically significant. Marinheiro (2001) estimated the three parameters for Portugal as (α,μ,λ) = (0.793,0.018,0.337) but such high propensity to consume out of total wealth is economically implausible. Lucke (1999) reported a fraction of household income facing liquidity constraints to be around 0.1.

Norway: Selected Issues
Author: International Monetary Fund