A Co-integration Approach to the Effect of Oil Income on Resource Allocation in Iranian Economy

This paper investigates the effect of oil income on real exchange rate defined as relative prices of non-tradable to tradable sector in Iranian economy. Increase in oil income, increases demand for both tradable and no-tradable goods. Tradable goods prices follow international prices while non-tradable prices are set in domestic markets. Therefore increase in oil income will result in real exchange rate appreciation and change in resource allocation in different economic sectors. Oil windfalls have changed the structure of the economy and relative prices such that the shares of agriculture and industry have decreased and the shares of services and construction have increased in GDP. The results show that there is a long run co-integration relation between oil income, capital outflow, GDP and real exchange rate.

the paper constructed as follow.Second section reviews oil income effects on resource allocation in the Booming Sector tradition.The third section uses a co-integration approach to test positive effect of oil income on real exchange rate appreciation.Section four concludes the remarks.

Booming Sector and Resource Allocation
There exists a huge literature on why countries might suffer "resource curse" (Auty,1999(Auty, ,2001)).Six transmission mechanisms can be considered-a long-term decline in terms of trade; revenue volatility; "booming Sector theories"; crowding out effects; increasing the role of the state; and finally the socio-cultural and political impacts."Booming sector theories" focus on the impact of a booming sector on resource allocation in the rest of the economy.The most widespread theoretical explanation of this apparent puzzle is found in models of the Dutch disease, where resource abundance shifts factors of production away from tradable sectors.Studies by Van Wijnbergen (1984), Krugman (1987), Matsuyama (1992), Sachs and Warner (1995) and Gylfason et al. (1999) finding show that the exploitation of more natural resources shrinks the traded ( industrial and agricultural) sectors, and thus total factor productivity across the economy reduces.This literature has been most influential in explaining why resource wealth may lower growth.
Positive oil price socks increase real income.These shocks have two main effects; expenditure effect and factor movement effect.In the resource movement effect, the resource boom will increase factor productivity in leading sector which increases the demand for labor.Factor productivity growth causes factor of production to shift toward the booming sector away from the lagging sectors i.e. industry and agriculture.This shift in factors from the lagging sector to the booming sector is called direct de-industrialization and direct de-aggriculturalization.However, this effect can be negligible, since the hydrocarbon and mineral sectors generally are capital intensive.Spending effect which is called indirect-deindustrialization and de-aggriculturalization occurs as a result of the extra revenue brought in by the resource boom.It shifts factors of production away from the tradable sector into non-tradable sector.As a result of the increased real income, demand both tradable and non-tradable shift up.While the prices in the traded good sector follow the rule of one price, and set internationally, non-tradable price and demand determined in domestic markets.So as a result of real income increase, non-tradable prices increase in terms of tradable prices.This profitability increase of non-tradable sector, translated into real exchange rate appreciation.Facing an appreciated exchange rate, resources flow to the more profitable productive sectors, that is the booming sector and non tradable, and demand shifts to inexpensive imports.This is what is known as the "Dutch disease," in reference to appreciation of the real exchange rate in the Netherlands when it started to export large quantities of North Sea natural gas (Cordon,1984).
In many cases, the sector most affected by this changing price structure has been agriculture.Industrial sectors have often been protected by tariffs and import restrictions in partial compensation for appreciation of the real exchange rate.Again, this equilibrium reflects the new comparative advantage of the country producing the booming commodity, yet it implies important restructuring of the whole economy.Furthermore, if, as in the case of oil, the sudden price boom is of short duration, economies find themselves with dismantled tradable sectors at the onset of the downturn of the Dutch disease.Only a few farsighted countries, like Cameroon or Botswana (Hill, 1991), protected themselves against such costly fluctuations by sterilizing foreign earnings abroad and regulating their entry into the country.This avoided an appreciation in the real exchange rate and saved foreign resources for times beyond the short commodity boom.Other mechanisms that can be used to prevent appreciation of the real exchange rate include increased government reserves out of taxation of foreign revenues, repayment of foreign debt, and exports of capital for direct foreign investment in other countries.

3.1.Real exchange Rate : Alternative Definitions
Purchasing power parity theory defines the real exchange rate as ; where p and p f are domestic and foreign price indexes , and e is the nominal exchange rate, defined as the domestic value of a unit of foreign currency.Note 1) The choice of nominal exchange rate to construct real exchange rate is controversial.Pinto(1990) makes a theoretical argument that, when the official market is rationed and domestic currency is only convertible in the floating parallel market at the free rate, imports purchased with official dollars are priced in domestic currency at their opportunity cost which is the parallel market nominal exchange rate.In the next section we show that the most appropriate nominal exchange rate that can indicate change in relative prices (the appropriate nominal exchange rate for construction of real exchange rate) is parallel market nominal exchange rate.Bahmani-Oskooee (1993, p.4055) argues that the black market nominal exchange rate served as an indicator of present and future price of imported as well as domestically produced goods.Whenever the dollar was up, so were the prices of all tradable and non tradable goods.
Alternatively, the real exchange rate can be defined as the relative price of non tradable in terms of tradable goods.Although these two definitions differ, the former is used as the proxy for later in most empirical studies (Edwards, 1989, pp.1-8) (Note 2).Then real exchange rate, defined as , where p N and p T are non-tradable and tradable price indexes respectively defined as the ratio of value added of each sector in current prices to the value added in real (constant) prices(Note 3).To construct RER both tradable and non tradable sectors are decomposed into consumption and capital goods.Tradable capital goods include machinery and equipment and non tradable capital goods includes construction.Value added in service sector is treated as non tradable consumption goods.Services include public, social, professional, monetary and financial services, transportation and communication, hotels, restaurant and commercial activities.They include governmental services as well as.This concept simply shows in relative prices and competitiveness of domestic tradable goods against foreign importable goods.Although the two definitions differ, the former is used as a proxy for the later in empirical studies.(Note 4) The purchasing-power parity hypothesis in its simplest form suggests that the real exchange rate is constant at the level reached at a time of macroeconomic balance and that reversions to the mean from any observed deviations are rapid.However, empirical evidence in support of this hypothesis is limited.To explain this discrepancy, several recent studies have examined time-varying sources of fluctuation in equilibrium real exchange rates as an alternative to the purchasing-power parity hypothesis.(Note 5) MacDonald and Ricci (2003) and Koranchelian (2005) study the effect of resource endowments (e.g., oil discoveries), terms of trade (e.g., oil prices), real interest rates and labor productivity differentials relative to a country's trading partners, and changes that arise as a result of economic policies and other factors, but do not examine the effects of parallel market rates on the equilibrium real exchange rates.Our assertion is that, when prices are flexible and the nominal exchange rate is frequently floating, the RER depends on the real fundamentals of the economy.Changes in real fundamentals such as log-term productivity growth, the underling capital flow, the term of trade and world economic conditions cause change in the underling demand and supply in real sector and consequently lead to change in real exchange rate.Against the proponents of purchasing power parity that believe equilibrium real exchange rate is constant, we show that it depends upon fundamentals that may change over time.

A co-integration approach to the effect of oil income on real exchange rate fluctuations
In regressing a time series variable on another time series variable(s), one often obtains a very high R2 even though there is no meaningful relationship between the two variables.Sometimes we expect no relationship between two variables, yet a regression of one on the other variable often shows a significant relationship.This situation exemplifies the problem of spurious, or nonsense, regression.Empirical works based on time series data assume that the underlying time series are stationary.Broadly speaking, a stochastic process is said to be stationary if its mean and variance are constant over time and the value of the covariance between the two time periods depends only on the distance or gap or lag between the two time periods and not the actual time at which the covariance is computed.In the time series literature, such a stochastic process is known as a weakly stationary, or covariance stationary, or second-order stationary, or wide sense, stochastic proce.(Banerjee and etal, 1993).Using non-stationary time series in a regression model lead to what Granger and Newbold (1974) called a spurious regression, and in this case statistical inference is not reliable.Therefore before estimating the model parameters, we test the existence of unit root in model's variables.Augmented Dickey-Fuller (ADF) and Phllips-Perron(PP) simple and joint tests were used to determine whether time series are stationary or not.Results for ADF and PP tests are reported in table5.1.The results show that none of ADF and PP simple and joint test can strongly reject the null hypothesis of unit root in model's variables.{Table 5.1} Most of non stationary economic time series are difference stationary.To test whether model's variable are trend stationary or difference stationary, we test first difference of model's variable.ADF and PP simple and joint test shows that first difference of all variables of this paper is stationary.The results are reported in table 5.2.The results show that at three interval confidence, unit root of first difference of variable strongly is rejected.{Table 5.2} Since all variable are integrated of degree one, they can be used in a regression model without fear of spurious regression.This paper uses two-step Engle-Granger co-integration test to show that there is a long run relationship between oil revenue, real GDP and capital outflow, and real exchange rate movement.In this method, first it estimates model's parameters and then checks residuals of model for unit root.If model's residual are stationary, estimated model is a non-spurious regression and there is long run co-integrating vector between model's variables.
In the case of Iran, analysis of real exchange rate and oil shocks has been conducted from various perspectives.Using official nominal exchange rate and the consumer price index, Pesaran (1992 P.113-115) constructed indices of bilateral and effective real exchange rates for the 1960-89 period.He argued that both indecies show real appreciation after the revolution, inconsistent with the changes in fundamentals of economy .Bahmani-Oskooee (1993) used the consumer price index and the parallel market nominal exchange rate, and showed that real exchange rate was constant during 1973-1986.We hypothesize that RER movements depend on economic fundamentals.Our model to investigate RER fluctuation is as follow , where, OIL stands for oil income in US dollar, GDP is real gross domestic product in real prices, and CO stands for capital outflow in real prices(Note 6) (Note 7) According to this model, a positive oil shock brings extra purchasing power to both public and private sector and as a result the demand for both tradable and non-tradable goods increases.The price of tradable goods id determined on world markets.Therefore, only the price of non-tradable goods will increase to restore the equilibrium in the real sector.This direct effect of oil revenue shock on RER which is called the real spending effect has different implication for public and private sectors, due to the fact that the positive oil export shocks has a huge impact on government consumption of non-tradables.Then we expect that  2 takes positive sign.Change in the domestic real income change the underling supply and demand of goods and services and consequently move the real sector of the economy out of equilibrium at the initial RER.To restore the equilibrium, the RER must adjust to the new conditions.Hence real income may have positive or negative effect.An exogenous increase in capital outflow necessitates a real exchange rate depreciation to increase equivalently the current account surplus and hence restore equilibrium in real sector.Then we expect  3 to take negative sign.
Table 5.3 show the result for a simple unilateral regression based on assumption that there is just one co-integrating vector between models variable.Based on t-statistics null hypothesis of each single coefficient can not be rejected and using F-Statistics, regression model is totally significant.{Table 5.3} There is a co-integrating vector between model's variables if model's residuals are stationary.ADF and PP are used to test stationarity of residuals.Both simple and joint test at three critical levels reject strongly existence of unit root in residuals, then model's residuals are stationary and hence there is one single co-integrating vector between models variables.Engle-Granger (1987) co-integration method assumption is that there is only one single co-integrating vector between variables of model.In other word it is based on assessing whether single-equation estimates of equilibrium error to be stationary.The second approach to co-integration, due to Johansen (1988,1991) and Stock and Watson(1988), is based on the VAR approach.The second method is based on the assumption that, between M variables, there may be more than one co-integrating vector.On this assumption first we should test for rank of VAR and then estimate long-run parameters.Co-integration test shows that there is one co-integrating vector between model's variable.Table4,5 show the result for co-integration test.Based on  max test rank of VAR is one and there is one co-integrating vector between model's variables.Using maximum likelihood method, we estimate long run parameters.Normalized parameters are reported in table 5.The results show that estimated parameters of two methods are the same, and consequently there is one long run relation ship between RER, oil income, real GDP and capital outflow.{Table 5.4} {Table 5.5} In general, the results show that 90 percent of relative price movement in Iranian economy can be explained by oil revenue, capital out flow and real GDP.Also results suggest that oil revenue is the main source of resource allocation in Iranian economy.Contrary to Karshenas (1994) that assumes there is no room for Dutch disease hypothesis to explain resource allocation in Iranian economy, our results show that Dutch disease hypothesis is strongly proved in Iranian economy and beside of institutional factors, change in relative prices has reshaped resource allocation.

Concluding Remarks
Over the past decades, developing countries that are rich in natural resources have performed significantly less well in economic terms than those that are resource poor.According to existing literature, resource abundance lowers economic growth.Dutch disease models describe negative effect of oil revenues on resource allocation in exporting countries.During last decades oil has have a dominant role in Iranian economy.Oil income had reshaped the whole economy and changed the role of public sector after it became the main source of foreign exchange and public revenue.Using co-integration approach we showed that oil revenues determined relative prices in Iranian economy.The results show that contrary to purchasing power parity , real exchange rate not only is not constant, but also changing fundamentals such as oil income, capital outflow, that change underling demand and supply in real sector, change relative prices.
Published by Canadian Center of Science and Education 199 Note 5. See Hinkle and Montiel (1999) for a survey of the challenges in assessing a country's equilibrium real exchange rate Note 6. Williamson (1986) argues that the current account surplus is the mirror image of underlying capital outflow and appropriate for the estimation of effects on real exchange rate.Note 7. The base year is 2000, and we use GDP deflator to transform time series into real term.

Table 5 .
1. Unit Root Test for the Level of Variables

Table 5 .
2.Unit Root Test for the First Difference of Variables