Israel-America Chamber of Commerce

 

The Link between Exchange Rate Uncertainty and Israeli Exports to the US¹

Dr. Yaron Zelekha²

 

Abstract

 

Israel is a small and open economy with a high proportion of foreign trade activity relative to output in general and high technology export in particular. Exports to the US, Israel’s largest trading partner, are prominent in this context. Therefore, it is important to understand the variables that determine Israeli exports, as well as the sensitivity of US imports to fluctuations in the exchange rate of the US dollar against the currency of the exporting country.

This study presents a brief description of a theoretical model for testing the effect of export price uncertainty on exports. Using this model, the effect of uncertainty on Israeli exports of goods and services to the US will be tested empirically. The estimation uses a quarterly sample for the period 1995–2007.

The results of the estimation and the model’s relatively high explanatory power support the research hypothesis that export price uncertainty has a negative and dominant effect, which is statistically significant, on Israel’s exports to the US. It was also found, as expected, that the main variables affecting exports in the long run are supply factors (including traditional variables such as relative price, the workforce and local productivity). Therefore, as mentioned, in one part of the literature (including the theoretical model developed in this study) uncertainty is described as a variable with a long-run effect.  

 

1. Background

The lack of certainty regarding economic variables that influence production is a problem that characterizes the productive sector in general and is discussed in the literature both on the level of the firm and on the level of aggregate investment. The path-breaking article by Hartman (1972) tested the effect of uncertainty on the firm’s production decision. The article related to both uncertainty regarding the future price of output to be received by the firm and uncertainty regarding labor costs and investment that will be required in the future. Since then, interest has grown in the effect of uncertainty (of various types: ranging from economy-wide uncertainty to price uncertainty and industry-wide shocks) on various components of demand and in particular on private consumption and investment. Studies have been carried out under various assumptions regarding the degree of risk aversion among individuals and firms, i.e. under the assumption that individuals and firms are risk averse or alternatively that they are risk indifferent. Within the framework of these studies and theoretical research into the link between uncertainty

and private investment, attention has also been given to the question of private investment by exporting firms on the level of the individual firm, as well as the effect of uncertainty on the activity of the export sector as a whole. Export price uncertainty is manifested in uncertainty regarding the relative value of export revenue, i.e. export prices in local terms.

In this context, De Grauwe (1988) claimed that income effects—which dominate substitution effects—are likely to result in a positive effect for price uncertainty, including exchange rate uncertainty, on the quantity of exports. Substitution effects are a result of the preference to shift production to the local market in the case of a risk of a low exchange rate while income effects are a result of the desire to increase the quantity of exports in order to compensate for the possible erosion in the exchange rate. He claimed that if exporters are characterized by sufficient risk aversion, then increased volatility in the exchange rate will raise the expected marginal utility of export revenue. Therefore, there will be a greater incentive for firms to increase their exports in a way that will overcome the substitution effects. Alternatively, exporters who have low risk aversion will not be concerned with the possible drop in export revenue due to exchange rate risk and therefore the income effect for them will not be as dominant.

Caballero and Corbo (1989) also point out that the effect of exchange rate uncertainty on exports is related to questions regarding the intensity of risk aversion. They claim that if risk aversion is assumed and if aggregate activity is correlated with innovation in world trade, then the link between exchange rate uncertainty and exports is liable to be negative. On the other hand, if the level of risk aversion is not high, then this negative effect will not outweigh the positive effect resulting from the convexity of the profit function relative to prices.

Dixit (1989) believes that the costs of adjustment resulting from the shift to or from exports are significant with respect to the nature of the firms’ reaction to exchange rate risk. These costs create a phenomenon of hysteresis or inertia in export flows. Higher uncertainty will lead firms to refrain from investment in export activity while on the other hand firms that are already producing for export will refrain from exiting (since they are unable to sell off past investment).

Caballero (1991) shows that the differences between the various models, which produce contradictory results with respect to the effect of exchange rate uncertainty on exports, are a result of a number of factors, some of which operate on the micro level and others on the macro level. On the micro level, these include asymmetry in the adjustment costs of production lines in models that show a negative effect (in other words, adjustment to lower production is more expensive than adjustment to an increase in production of the same magnitude) in contrast to symmetry in other models which show a positive effect and declining or fixed returns to scale in the production function. On the macro level, the differences in the effect of exchange rate uncertainty on the quantity of exports is a result of the different basic assumptions made regarding the structure of competition, i.e. the degree of competition in the market. Therefore, whether the effect is positive or negative is a function of the degree of competition in the markets and the nature of the production function, as well as the interaction between the two. The more competitive markets are and the closer returns to scale are to being fixed, the greater the positive effect of uncertainty will generally be on investment. Caballero concludes that the negative effect of uncertainty from various sources is more common than the positive effect.

It is worth mentioning in this context that the approach in which external shocks to investment in the economy are the result of a mixture of micro factors (which relate to the nature of the production function) and macro factors (which relate to issues of equilibrium in the markets and their degree of competition) is the accepted one in the literature. See, for example, Bachmann, Caballero and Engel (2006).

Pindyck and Caballero (1996) show that on the level of the economy as a whole the effect of uncertainty (in a firm-level equation) is negative even when adjustment costs are symmetric. The explanation for this lies in the difference between the effect of a positive demand shock on the entry of new competitors (who are waiting for the increase in prices resulting from the shock) and the smaller effect of a negative shock on the exit of competitors. Therefore, on the level of the firm, uncertainty (which is not related to the entry or exit of competitors from the market) has less of an effect than uncertainty on the level of the economy.

Tavlas and Swamy (1997) claim that as exchange rate uncertainty increases, so does the accumulation of knowledge with respect to exchange rates among exporting firms. This expertise is likely to create an additional source of profits for them, which will contribute to the positive effect of uncertainty on exports on the level of the economy.

In addition to the many theoretical studies, a great deal of empirical work has attempted to test for the effect of export price uncertainty and exchange rate uncertainty on exports.

In this context, it is worth mentioning a number of studies, including Caballero and Corbo (1989) who found, based on data for six developing countries, that exchange rate uncertainty has large negative effects on exports. Moreover, they found that the long-run effect of uncertainty is even larger than that found in the short run and they explain this through risk aversion. According to them, in every situation where uncertainty is in fact realized the firm must absorb a loss and this is the source of its risk aversion. This is true whether the quantity of its investment turns out to be too large relative to revenues in terms of the local currency or whether it turns out to be too small (i.e. an alternative loss).

Similar results, both in the long and short run, were arrived at by Arize, Osang and Siottje (2008) who used data for eight Latin American countries over an extended period of thirty years. In some of the countries, it was found that uncertainty has a larger effect than the relative prices of tradable and nontradable goods. In their opinion, the results show that exchange rate uncertainty has a negative effect on exports also in countries with medium per capita income and not only low per capita income. They also emphasize the importance of testing for the effect of particularly large uncertainty in countries that have moved from a fixed exchange rate to a flexible one, as a result of the generally greater volatility under a flexible exchange rate regime. In these countries in particular one should be careful to include uncertainty in estimation since ignoring it is liable to lead to misspecification and the resulting distortion of results. In this context, Bahmani-Oskooee (2002) found a negative effect for developing countries outside of Latin America as well.

In contrast, findings for developed countries are not unambiguous and in fact show some support for a positive effect of real exchange rate uncertainty on exports. This contrasts with studies of developing countries which usually show large negative effects. While Arize (1995) and Choudhry (2005) report negative effects, Qian and Varangis (1994) and Baum, Caglayan and Ozkan (2004) report a negative effect in only some of the selected countries and a positive one in the rest. Meanwhile, Doyle (2001) reported positive effects in bilateral trade between Britain and Ireland in most industries (and negative effects in the rest). Chourdhry (2008) reported statistically significant positive effects in trade between Britain on the one hand and Canda, New Zealand and Japan on the other, both in the short and long runs, as do McKenzie and Brooks (1997) and others. In this context, Doyle points out that the positive effects are related to multinational ownership of companies involved in foreign trade. Multinational ownership provides natural protection against some of the movements in the exchange rate through the ability to shift production or exports between subsidiaries in different countries.

The results in the literature therefore seem to indicate a direct link between the stage of a country's development (and particularly of its financial markets which provide the ability to protect against unexpected changes in exchange rates to some extent) and the sign of the effect of uncertainty on trade. The various studies also indicate that results are dependent on the statistical method used (see below) and the chosen sample period. Thus, in broad terms, a negative effect has been found in developing countries and a positive effect in some of the studies of developed countries.

It should be mentioned that another important factor related to a country's level of development is the proportion of added value in export goods. The higher is the proportion of added value, the easier it is for exporters to accumulate greater market power and thus to raise the price of their products. At the same time, this will make it harder for an exporter to shift his products to the more limited local market. Furthermore, a high proportion of added value is likely to be related to high fixed costs (primarily R&D) and therefore a high rate of contribution-based costing (i.e. pricing per unit of product which takes into account only variable costs). In this situation, there is a greater ability to absorb a drop in revenue in terms of the local currency since the effect of small changes will be no higher than the loss in production of the marginal unit. Another explanation may be related to the ability to achieve greater efficiency. It is reasonable to assume that producers of goods with a high proportion of added value will have greater expertise in streamlining the production process (i.e. increasing productivity) in order to compensate for changes in total revenue. Similarly, high fixed costs affect the size of adjustment costs involved in the entry and exit of firms (which in the literature constitute a significant component in the total effect on exports). Finally, the level of a country's development is likely to be related to the globalization process among firms operating in that country. In general, multinational corporations (i.e. companies that produce in a number of countries and export to a number of markets) find it is easier to shift production and marketing resources in a way that will minimize changes in revenue in terms of the local currency, whether they are the result of a change in the specific exchange rate that works against them or whether they are the result of a change in export prices in terms of the export currency (which provides natural protection that does not have to be purchased in the capital market).

However, the results in the literature are not unambiguous and therefore an empirical estimation is required for each and every economy. Furthermore, Barkoulas, Baum and Caglayan (2002) even suggest that the various types of uncertainty are, among others, the factors behind the contradictory empirical results. In their view, while one type of uncertainty (such as, for example, the structural uncertainty in the local capital market) leads to a particular effect, a different type of uncertainty (such as policy uncertainty) leads to the opposite effect. It should be mentioned that this conclusion in the literature is also relevant in distinguishing between uncertainty resulting from a change in export prices in terms of a fixed exchange rate and uncertainty resulting from changes in the exchange rate while the price remains fixed in terms of the export currency.

Indeed, despite the importance of the issue and even though the exclusion of important variables is liable to distort the results in general and the reported effects of other parameters in particular, empirical studies of the Israeli economy do not relate to the effect of export price uncertainty (in terms of the local currency) on exports.

The article is composed of five parts: Section 2 will develop a theoretical model that is structured to include the effect of export price uncertainty on Israeli exports (as in the literature, the analysis will focus on the value of export revenues in terms of the local currency). Section 3 will describe the database and the statistical problems in its use. Section 4 will present the results and Section 5 contains a summary and conclusions stemming from the results, including policy implications.

 

2. The theoretical model

Despite the many theoretical and empirical studies conducted worldwide since the breakthrough by Hartman (1972), the effect of uncertainty in general, and the effect of price uncertainty on private investment in particular and therefore on exports, has remained an open question on a theoretical level with insufficient empirical proof to resolve it one way or  the other. Meanwhile, in Israel, the issue has not been studied at all, despite the liberalization of capital flows and the dependence of the economy on foreign trade and despite the major changes that have characterized the exchange rate regime since the Stabilization Plan in 1985 (which involved a transition from a fixed exchange rate regime to a semi-floating regime—a crawling band that was adjusted a number of times over the years—and finally to a floating regime in recent years).

In any case, for an economy with firms whose production functions are characterized by constant elasticity of substitution between factors of production (i.e. labor and capital) the first-order conditions for profit maximization require that the optimal stock of capital be determined in the following manner:

 

(1)                                               

 

where α is the production elasticity of capital, γ is the elasticity of substation between capital and labor, P_n is the stochastic distribution of the expectations of relative prices and of exchange rates (which is a function of the probabilities for the rates of real change in export prices in terms of the export currency and possible exchange rates ), C is the price of capital services and y is business output.

The neoclassical model is based on the Cobb-Douglas production function which requires that γ=1. The process of adjustment of the actual capital stock to its optimal level can be described by the following equation:

 

(2)                                                    

 

where

IN - net investment by the business sector less physical deterioration.

β – coefficient of adjustment.

 

Within this framework, β=1 if adjustment is carried out in full within one period and  if the adjustment process occurs over a number of periods.

By substituting (1) into (2), we obtain an equation for net investment that is compatible both with the neoclassical approach and the rational expectations theory, which is modified (as described above) to include export price uncertainty in real local terms as follows:

 

(3)                                               

 

For gross investment, the effect of physical deterioration is added. Thus:

 

(4)                                           

 

where:

I – gross investment by the business sector.

δ – rate of physical deterioration of the capital stock.

 

Since investment for purposes of export must equal export revenues in the long run, we obtain:

 

(5)                                          

 

where:

X – exports.

 

It should be mentioned that an alternative approach can be taken by making a somewhat artificial modification of De Gregorio and Wolf (1994)’s model which is a good approximation of the small and open Israeli economy. The basic assumptions of this model are a balanced current account and government deficit and therefore it is particularly suited to a long-run analysis, in which according to the theory the only explanatory factors are related to supply and the terms of trade. De Gregorio and Wolf’s model can be artificially expanded through the addition of various cost components to the cost of labor and capital in the production function of the export good. Thus, we will claim that an expansion can be performed with respect to uncertainty and the related cost of protection. It should be stated in this context that the approach is also consistent with the empirical findings reported for other countries, which showed that the more developed a country is, the greater is its ability to obtain protection at a reasonable price from unexpected changes in the local exchange rate. In other words, the cost of protection is a negative function of a country's level of development. Therefore, the addition of a permanent cost component representing uncertainty is in line with the accepted theory since it weights the cost of protection within a firm’s production function and therefore also exists in the long run. The main disadvantage of the model is the way in which uncertainty enters, i.e. as an exogenous factor that does not develop as an inherent and structured part of the model itself.

 

3. The database and the statistical method

The model developed above can be tested empirically in order to obtain an estimate of the effect of the explanatory variables on exports and particularly the effect of export price uncertainty in real local terms. However, the empirical analysis involves a number of complex statistical problems that need to be clarified beforehand.

The first problem is related to the composition of the basket of currencies according to which Israel’s export prices are weighted. Although the Central Bureau of Statistics publishes a weighted index of export prices, the method of weighting is liable to degrade the quality of the data (in a similar manner to seasonal adjustment) and therefore their use makes it problematic to use specific exchange rates and the uncertainty attached to them, which is essential for dealing with a different statistical problem to be described below. Therefore, as in most of the studies done worldwide, we will carry out an empirical estimation for Israel’s exports to a specific country, i.e. the US, which has been Israel’s largest trading partner over many decades, using both export prices and the real exchange rate of the US dollar against the shekel (deflated by the index of output prices, which was chosen because it represents the profitability of investment in the expansion of production lines better than the CPI does).

The second problem is that of simultaneity which exists between the exchange rate and exports. The two influence each other in a way that distorts the standard statistical treatment in a simple regression. Some studies used lagged variables as part of co integration estimation but in this way dealt only partially with the problem. Alternatively, the problem can be dealt with comprehensively by using two-stage estimation.

In the first stage, the real exchange rate variable (i.e. the nominal dollar/shekel exchange rate deflated by the index of output prices in Israel) is estimated using the instrumental variables method. The equation will include total productivity variables for Israeli industry and total productivity in US industry (in the absence of data on production in tradable goods industries, the two variables together constitute an estimator of the former), total productivity in the non-tradable goods industries (an increase in which expands the supply of nontradable goods and reduces their price in a way that contributes to a real depreciation), Israel’s terms of trade, per capita output as an expression of individuals’ purchasing power, the import surplus as a percentage of output which represents the intensity of local demand pressure on sources, the rate of investment in the economy as a percentage of output, the per capita stock of capital in the business sector or alternatively per capita investment, labor-intensive public consumption as a percentage of output (on the assumption that it competes with the business sector with respect to the labor force and its wages) and an estimator of Israel’s security situation (using the number of Israelis killed each quarter).

In the second stage, Israeli exports to the US are estimated on the basis of an equation that includes separate estimators for two types of export price uncertainty in local terms: that related to the exchange rate (see below) and that related to dollar export prices. In other words, the equation will include the real exchange rate estimated in the first stage, export prices (relative to local output prices), the size of the population (aged 15 and over) as representative of the economy's potential workforce, the stock of capital in the business sector, the productivity of industry in the US and in Israel (representing technological innovation), total US imports as representing the demand for Israeli exports, one-year interest rates in the US and in Israel and finally the security situation variable in view of both its direct effect on the export of tourist services and its indirect effect on the uncertainty among buyers abroad (with regard to exporters meeting delivery schedules) and their willingness to buy from Israel.

The third problem relates to the estimators of price uncertainty themselves. The literature describes three methods for including uncertainty which are intended to deal with the problem of serial autocorrelation in the price and exchange rate standard deviation data (see Jansen, 1989). The serial autocorrelation is a result of the inertia of external noise within daily currency prices which do not adjust immediately but rather continue to have an effect over time (particularly white noise related to expectations of policy changes or the analysis of their effect or, in the case of Israel, to security developments). The first and most common method in the literature calls for the inclusion of moving averages of the standard deviations over an extended period of time. The disadvantage of this method lies in the incomplete elimination of the serial autocorrelation. The second method uses the GARCH model in which the standard deviation is weighted in a formula that derives the trend of the standard deviation from the raw data. The main disadvantage of this method is the excessive elimination of secondary trends in the standard deviations which is likely to neutralize the overall trend and to seriously distort the fluctuations in the data. It should be mentioned that according to Klaassen (2004) the first two methods may even create contradictory estimates of the trend in risk and uncertainty. The third method, which is recommended by Baum, Caglayan and Ozkan (2004) (and previously by others) and which we will use alongside the first method, is to use the calculated standard deviations of the expected exchange rate in the form of one-month-ahead spot prices. This solution also requires the estimation of exports to only one country rather than total exports, as mentioned above.

The fourth problem is a result of unit roots in the variables and therefore, before choosing the method of statistical estimation, a test for stationarity (the unit root problem) was performed on the variables using an Augmented Dickey-Fuller (ADF) test. This test involves the estimation of Equation (6) in order to obtain the first-order differences of each variable x_t:

 

(6)                                     x_t – x_t-1 =  α + βx_t-1+ U_t

 

Since unit roots were found in the errors of the equations, the hypothesis was tested that β is equal to zero for each equation separately. It was found that some of the variables are not stationary and that their variance is infinite. At the same time, a parallel test that was done on the first-order differences of those same variables showed that they were all stationary at an acceptable level of confidence, apart from the population variable. It should be mentioned that the large wave of aliyah gradually decreased in size over the sample period and as a result significant changes occurred in the rate of population increase. As a result, the test for stationarity did not show convergence, even in the first-order differences. However, there is little doubt that in the long run the population converges and this is certainly true for the first differences (and this is indeed seen in the tests for stationarity performed on the first differences of population data for other periods) and therefore this variable can be included in the long-run equations. Alternatively, the population variable was replaced by a trend variable (which receives an increasing numerical value for each period) in a way that does not compromise the quality of the estimation (see below). In any case, the results of the test indicate the possibility that the variables are integrated of first order. In this case, it is possible that an co integration equation can be created in which the composition of variables is stationary, on the condition that the errors in the equation do not reveal a unit root. Ramanathan (1998) recommends an ADF test for the errors and indeed this test showed that the equations create first-order integration. Therefore, a t-test can validly be done on them and OLS can be performed on the original variables. It should be mentioned that the stationarity of the error is meant to express a long-run relationship between the dependent variable and the explanatory variables. However, since the sample is composed of quarterly observations that are relatively short run in nature, the interpretation adopted in the present study views these results as also reflecting intermediate-run trends. This interpretation is also in line with the conventional wisdom in Israel that the building of the Israeli economy continued for decades (and is still continuing) and therefore short-run dynamics are likely to play a major role over very long periods.

 

4. Results

The various versions the equation explaining Israeli exports to the US (see Table 1 for the results) produced coefficients of correlation of up to = 0.980 (R² =0.983) with SE=0.055 and DW=1.809 (which indicate the absence of serial correlation). Most of the important variables were found to be significant at all the accepted confidence levels (99% and even higher) and all had the expected sign. In addition, the estimates had relatively low variances.

The relative price variable was represented in all the versions by the index of export prices (adjusted for the change in output prices) and its elasticity was found to be positive and close to one (0.856) as expected.

At the same time, it should be mentioned that the real exchange rate variable, which was estimated in the first stage of the two-stage estimation in an equation with a high level of significance ( =0.816), was not found to be significant. Moreover, although the sign of the coefficient in all the versions was positive as expected, in some of the versions the export prices variable had a low level of significance (about 87% while the elasticity of the coefficient reached a level of 0.452). It is possible that a higher level of significance in the first-stage estimation would have made it possible to achieve an acceptable level of significance in the second stage of estimation.

The productivity variable was represented in all the versions by productivity in Israeli industry while US productivity was not statistically significant. It became evident that technological innovations are better represented by productivity in Israel. The elasticity of productivity in Israeli industry reached a level of 0.13.

The labor force variable was found to be significant at the highest level of confidence in all the versions and its elasticity reached a level of 3.997. This result is explained by the growth in population, which was characterized by a constant upward trend throughout the sample period. When we chose to include a trend variable instead of population, the equation remained unchanged. In other words, population behaved like a trend variable while the other variables remained stable. It is worth mentioning that there was an expansion of exports during the sample period with fluctuations around an upward trend and it is possible that the array of factors that determine this trend (which is characterized by a time component) should include the globalization process and the increasing openness of the economy.

Among the interest rate variables found to be statistically significant were the short-run real interest rate in the US lagged by one period, whose elasticity reached a level of 0.025, and the short-run interest rate in Israel, whose elasticity reached a level of 0.018. It is possible that the interest rates represent the rates of growth in US and Israeli output respectively and the former obviously affects the demand for Israeli export goods (while the Israeli short-run interest rate is generally correlated with the US interest rate). It is possible that in this context the US interest rate is therefore more highly correlated with Israeli exports than the quantity of US imports relative to US output, which was found not to be significant in all the versions.

The security situation variable, which was represented by the number of Israelis killed in terrorist attacks each quarter was found to be significant though with a relatively small effect (as expected in a long-run equation) with an elasticity that reached 0.001. It is worth mentioning that the increase in terrorist attacks, which was an expression of the deterioration in the security situation, constitutes a negative shock to the economy, which, if it persists, reduces the price level of non-tradable goods or in other words creates a real depreciation, which is indeed what was concluded from the data.

The business capital stock variable was found not to be significant in any of the versions examined. It is possible that this was a result of the fact that most of the business capital stock that appears in the figures of the Central Bureau of Statistics in fact  belongs to the non-tradable goods sector (which includes transportation) and therefore is only weakly correlated with the capital stock of the tradable goods sector.

The group of possible variables for uncertainty was represented primarily by the standard deviation in relative prices of exports (defined as the ratio between export prices and output prices during 10 quarters, lagged by one period). The estimation shows that the standard deviation of the real exchange rate (the nominal exchange rate deflated by the index of output prices) had what seemed to be a negative effect; however, it was not statistically significant and therefore was not included in the reported results. It is possible that the effect of dollar price uncertainty during the sample period was greater than that of exchange rate uncertainty. This finding is more compatible with the long-run equation since the variation in the real exchange rate between countries essentially represents differences in productivity, which are represented separately in the equation.

In addition, it is worth mentioning that not only was the effect of uncertainty highly significant but it was also of a high magnitude. The elasticity of the coefficient reached    -7.032 and in the version that included a trend variable instead of the workforce variable it reached -8.606.

With respect to the two alternatives for estimating uncertainty, it was found that the method of moving averages (as mentioned above, the standard deviations variable over 10 quarters was the one found to be statistically significant) is the one that provides a statistically significant answer to the test hypothesis. In contrast, the variable based on the standard deviation implicit in options (i.e. the method recommended by Baum, Caglayan and Ozkan) was not found to be statistically significant. Nonetheless, the data from the capital market made it possible to produce a series for only part of the sample period (from 2002 to 2007) and therefore they were combined with the actual standard deviations of the representative exchange rate for the first part of the sample period (with the addition of a dummy that would represent the juncture point and attempt to control for its effect). This is based on the work done at the Bank of Israel (Benit and Shreiber, 2003) which showed that the expectations implicit in options during the relevant period are adaptive. It may be that this technical constraint compromised the quality of the data and that the adaptivity of the series magnified its effect relative to the actual standard deviations. In any case, it is worth mentioning that the versions which tested for the effect of the standard deviation variable based on options also found that the effect was negative; however, as mentioned, it was not statistically significant.

In order to improve the estimation of the relationship between uncertainty and exports in the long run, relative to the short-run relationship, an Error Correction Model (ECM) was estimated. According to Ramanathan, if the system of variables forms a first-order co integration equation, then there exists a variable that represents the correction of the error, which can be included in the two-stage estimation and will distinguish between the effects in the short run and those in long-run equilibrium. In the first stage, the co integration equation is estimated and in the second stage the first differences equation is estimated, which includes the error obtained in the first stage lagged by one period. This error represents the ECM factor that is meant to absorb the deviation from equilibrium in the short run and which causes the other coefficients in the equation to represent long-run effects.

The results of the estimation are presented in Table 1 below. It should be mentioned that the model illustrates the fact that the standard deviation of expectations variable, which is used as an estimator of expected inflation uncertainty, has an effect both in the short and long runs. The ECM factor was found to be negative as expected, and statistically significant, at a level of -0.450. It is worth mentioning that the negative effect of the ECM factor is consistent with the theory (which also predicts an absolute value of between 0 and 1). The significance of this result is that convergence to equilibrium is expected after a random deviation. In other words, following a period in which a deviation from equilibrium occurs in one direction, one can then expect a correction in the opposite direction and a return to equilibrium—in our case after about two and a quarter quarters.

 

5. Summary and conclusions

Israel has a small and open economy that is characterized by a high level of foreign trade relative to output. As a result, it is important to understand the variables that affect Israeli exports and in particular its exports to the US, its largest trading partner.

In this context, under the assumption that in the long run demand in foreign markets for the majority of Israeli export goods is completely elastic at a fixed price in foreign currency, the quantity of exports will be determined primarily by supply factors. And indeed most of the factors that were found in this study to affect Israeli exports to the US in the long run were supply factors (including the traditional variables of relative price, workforce and local productivity), as well as an uncertainty variable that in one part of the literature (including the theoretical model developed in this study) captures a long-run effect while in another part captures a short-run effect. In addition, it was found that the inclusion of variables reflecting uncertainty in the security situation and US and Israeli interest rates (which are usually correlated with one another and which are, in addition, significant short-run variables) contributes to the explanation of Israeli exports to the US. It is possible that this is explained by the special characteristics of the Israeli economy—whose development has been affected by an uncertain security situation since its creation—alongside the strong influence of the US economy on that of Israel.

The results of the estimation and the high explanatory power obtained support the research hypothesis that price uncertainty has a negative effect on exports, which is statistically significant and dominates the determination of Israeli exports to the US.

It is possible that the negative effect found here is evidence that, among other things, Israel has an "emerging" economy. In other words, the fact that the effect of uncertainty was found to be so significant is apparently also due to the development process in Israeli capital markets during the sample period, including the possibilities for purchasing protection against currency movements for long periods and for large amounts. The development of Israeli capital markets in recent years, although it improved the situation somewhat (only recently have conditions been created in which the government can carry out SWAP transactions of dollar-denominated debt into shekels in quantities of several hundred million dollars and for periods of several years), has not reached the point where long-run protection against foreign currency movements relative to the shekel can be obtained for large amounts and/or for long periods (and certainly this was not possible to a significant extent during the sample period).

The findings have important policy implications both for Israel and the US. With respect to Israel, the findings point to the importance of encouraging the development of Israeli multinational corporations that invest in a number of markets and thus are able to diversify risk. In this context, the trend in recent years to reduce barriers to Israeli investment abroad is encouraging. As a result, Israeli companies will be able to more broadly diversify risk and to more easily purchase protection against changes in export prices in local terms. In addition, the findings indicate the importance of a concerted and consistent policy to accelerate the development of financial markets in Israel in general and their globalization in particular. As part of this effort, emphasis should be placed on the development of liquid and stable markets that can offer derivative financial instruments of various types. With respect to the US, the findings strengthen the conclusions reached in the literature, according to which the worldwide globalization strategy, which is increasing the efficiency of financial markets in developing countries that trade with the US, is having a significant effect on the sensitivity of US imports.

 

 

References

1.       Arize, A.C., 1995, "The Effects of Exchange Rate Volatility on US Exports: an Empirical Investigation", Southern Economic Journal 62, 34-43.

2.      Arize, C., Osang, T. and D. Slottje, 2008, "Exchange Rate Volatility in Latin America and its Impact on Foreign Trade", International Review of Economics and Finance 17, 33-44.

3.      Bachmann, R., Caballero, J. and E. Engel, 2006, "Lumpy Investment in Dynamic General Equilibrium", Discussion Paper 1566, Yale University.

4.      Bahmani-Oskooee, M., 2002, "does Black Market Exchange Rate Volatility Deter the Trade Flows? Iranian Experience", Applied Economics 34, 2249-2255.

5.      Barkoulas, T., Baum, F. and M. Caglayan, 2002, "Exchange Rate Effects on the Volume and Variability of Trade Flows", Journal of International Money and Finance 21, 481-496.

6.      Baum, F,. Caglayan, M. and N. Ozkan, 2004, "Nonlinear Effects of Exchange Rate Volatility on the Volume of Bilateral Exports", Journal of Applied Econometrics 19, 1-23.

7.      Benita, G. and B. Shreiber, 2003, "On Implicit and Actual Standard Deviations and the Relation between them", Issues in Foreign Currency 3/03. (Hebrew) 

8.      Caballero, R.J., 1991, "On the sign of the Investment – Uncertainty Relationship", American Economic Review 81, 279-288.

9.      Caballero, R.J. and V. Corbo, 1989, "The Effect of Real Exchange Rate Uncertainty on Exports: Empirical Evidence", The World Bank Economic Review 3, 263-278.

10.  Caballero, R.J. and Pindyck, R.S., 1996, "Uncertainty, Investment, and Industry Evolution", International Economic Review 37 (3), 641-662.

11.  Choudhry, T., 2005, " Exchange Rate Volatility and the United States Exports: Evidence from Canada and Japan", Journal of the Japanese and International Economies 19(1), 51-71.

12.  Choudhry, T., 2008, "Exchange Rate Volatility and United Kingdom Trade: Evidence from Canada, Japan and New Zealand", Empirical Economics Accepted and not yet published.

13.  De Grauwe, P., 1988' "Exchange Rate Volatility and the Slowdown in Growth of International Trade", IMF Staff Papers 35, 63-84.

14.  De Gregorio J. and H. Wolf, 1994, "Terms of trade, Productivity and Demand Shocks and Asia-Pacific Real Exchange Rate", National Bureau of Economic Research Discussion Paper No. 6108.

15.  Dixit. A., 1989, "Entry and Exit Decisions under Uncertainty", Journal of Political Economy 97 (3), 620-638.

16.  Doyle, E., 2001, "Exchange Rate Volatility and Irish-UK Trade, 1979-1992", Applied Economics 33, 249-265.

17.  Hartman, R., 1972, "The Effects of price and Cost Uncertainty on Investment", Journal of Economic Theory 5, 258-266.

18.  Jansen, D.W., 1989, "Does Inflation Uncertainty Affects Output Growth? Further Evidence", The Federal Reserve Bank of St. Louis Review, July-August, 43-54.

19.  Klaassen, F., 2004, "Why is it so Difficult to Find an Effect of Exchange Rate Risk on Trade?", Journal of International Money and Finance 23(5), 817-839.

20.  McKenzie M. and R. Brooks, 1997, "Exchange Rate Volatility and International Trade on German-US Trade Flows", Journal of International Financial Markets Institutions and Money 7, 73-87.

21.  Qian, Y. and P. Varangis, 1994, "Does Exchange Rate Volatility Hinder Export Growth?", Empirical Economics 19,371-396.

22.  Ramanathan, R., 1998, Introductory Econometrics with Applications, The Dryden Press 4^th Edition.

23.  Romanov, D., 2003, "The Real Exchange Rate and the Balassa-Samuelson Hypothesis: an Appraisal of Israel's Case Since 1986", Discussion Paper No. 2003.09, Bank of Israel, Research Department.

24.  Tavlas, G.S. and P.A. Swamy, 1997, "Macroeconomic Policies and World Financial Integration", in M. U. Fratianni, D. Salvatore and H. Von Itazen (Eds.), Macroeconomic Policy in Open Economics, 248-280, Westport, Connecticut, USA: Greenwood Press.

Table 1: Estimation of equations explaining Israeli exports to the US for the period 1995–2007*

 

 

* The equations are in log-linear form and therefore the workforce and export prices variables are expressed in natural logarithms. t-values appear in parentheses.

 

** All the variables, apart from the ECM factor, are expressed as first-order differences.