Improving Australia's trade balance: A case study of agro-forest and fish products

The impact of the exchange rate (ER) on the trade balance (TB) has been discussed for many years. However, the issue has not been discussed in sufficient depth, especially in relation to the TB of agricultural products. This paper will gauge this impact on Australia, which has much potential for agro-based trade in the world market. We have applied the Bahmani-Oskooee and Hosny's approach of the linear autoregressive distributed lag model to estimate the Marshall – Lerner condition (MLC) regarding the trade of Australian agro-forest and fish (AFF) products with its five major partner countries. Quarterly data will be used for the period 1988Q1 – 2020Q4. Our findings support the MLC in case of the major share of Australian AFF trade. The implication is that if the market force depreciates Australian ER, the country's AFF TB will improve in the long-run.

Table 1 shows the trade picture of Australia with her five major trading partners.These five countries have constituted about 43% and 55% of Australian AFF imports and exports, respectively, in 2020.Since Australia is a land abundant country and AFF production needs intensive use of land resources, Australia has much scope to enlarge its contribution by AFF in its total trade share.By considering the above backdrop, this article takes the initiative to test the validity of the MLC between Australia versus her major five AFF products trading partners.
To that end, we present the literature review in Section 2, outline the models and econometric technique in Section 3, explain the results and, report the result summary in Section 4 and draw the conclusions in Section 5.

| LITERATURE REVIEW
Depreciation usually increases and decreases import and export prices respectively in terms of foreign currency.So foreigners see that exports are less expensive and the countrymen get that imports are more expensive.Thus it is widely believed that TB improves when depreciation takes the place of the ER of a country.However, later Marshall and Lerner have shown that TB change is mostly a phenomenon of elasticities of import and export, not only of ER.Thus, it is now an established fact that the MLC approach is one kind of rule of thumb that is used to predict whether RER depreciation improves the TB of a country.There have been many studies on this issue across countries, regions, and periods, some of those have already paid attention to the test of the ER sensitivity and status of the MLC in context of Australia.All of these studies are chronologically tabulated in Appendix C.
First, Arndt and Dorrance (1987) take steps to analyse the Australian J-curve, using a selfgenerated tabular approach for their paper.Based on their findings they opine that nobody can rule out of having J-curve effect on the current account (CA) balance of Australia.However, they conclude that not only ER but also the efficacy of MLC, the competitive power of Australian exporting goods, exogenous factors to change of terms of trade (TOT), and level of domestic or national aggregate spending are key factors to TB improvement.
However, in the next Felmingham (1988) reaches in a converse conclusion.He examines the impact of TOT change on Australian TB using an approach presented by Haynes and Stone (1983) calculating and subsequently constructing a table comprising series of Australian TB and TOT.He also defines TB by import-export ratio.Subsequently, he attempts to relate TB with the current and past values of TOT, and decides that there is no evidence of the Australian Jcurve.His findings have subsequently been discussed and economists suggest that one probable weakness of his method is that perhaps his calculated TOT may have had a very low correlation with the actual ER of Australia at that time.Depreciation can change the TOT in either direction depending on the product of elasticities of the exports demand and imports supply.So, a more effective method to examine the impact of depreciation on TB is to relate the TB directly to ER not with else what Felmingham (1988) has not done.
To overcome the weakness of Felmingham (1988)'s study, Karunaratne (1988) tries to investigate the link of Australian TB not only to its TOT, but also to REER.He also concludes that change in REER does not have any significant influence on the Australian CAB.His methodology is also later criticised by Bahmani-Oskooee and Baek (2015) who has considered that there could be strong multicollinearity between the TOT and REER when they are both included in the same model which was perhaps not understood and, hence, addressed by Karunaratne.None of the above three studies provides any concrete evidence of the nexus between Australian ER and TB in either the short-run, or the long-run.However, using quarterly data for the period of 1977Q1-1988Q4, Bahmani-Oskooee and Pourheydarian (1991) have got sufficient indication that J-curve notion is valid for Australia in the short-run.Furthermore, their results show that devaluation of Australian dollar (AUD) also improves the TB in the long-run.
Bahmani-Oskooee and Niroomand (1998) criticise all the above researches noting that they have not followed proper econometric techniques due to the nonadvancement of time series techniques until their study period.It is true that none of the above studies has tested unit root of the data used though all of them rely on the data with time series dimension.To overcome this fault they use Johansen co-integration technique for 30 countries time series data of different periods individually.Our country of interest, Australia, is also included in this research paper using the annual data for the period of .Addressing all the time series properties of the data this study finds econometrically significant evidence of MLC for Australia for the first time.
Gradually, more studies start to emerge for Australia.Mahmud, Ullah, and Yucel (2004) checks MLC for six developed countries, Australia, Germany, Japan, Norway, United Kingdom and the USA by using the nonparametric technique resorting kernel estimation approach to estimate import and export price elasticities to gauge the MLC.Their results suggest that MLC is valid only for Norway.
Bahmani-Oskooee, Goswami, and Talukdar (2005) further investigates the short-and longrun effect of REER depreciation on Australian aggregate TB with 23 trading partners on a bilateral basis.The study also uses the aggregate trading data for the USA, Japan, China, South Korea and Thailand,-the top five partners, they consider for this study where quarterly data over the 1973-2001 period and co-integration technique of autoregressive distributed lag (ARDL) bound testing and error correction method (ECM) are used.The results show that among the trading partners related to this study except South Korea no other four countries support the J-curve phenomenon for Australia.
Again, Bahmani-Oskooee and Wang (2007aWang ( , 2007b) have criticised all the above studies because they are conducted by the data either between Australia and the rest of the world or between Australia and its individual trading partners aggregate data bases.So, these studies are suffering from aggregation bias.To overcome this fault, Bahmani-Oskooee andWang (2007a, 2007b) have used disaggregated data between Australia and its second largest trading partnerthe USA.They test for the annual United States-Australia bilateral trading data for totally 108 industries for the period of 1962-2003 using ARDL bounds testing and ECM approach for co-integration.Out of this 108 categories of products only 23 were AFF commodities.In this research they were able to discover J-curve effects only for 8 AFF commodities out of 23.
Furthermore, Bahmani-Oskooee and Wang (2009) have conducted another research adding two more years' data for the period of 1962-2005.In this paper they again use disaggregate trade data between Australia and the USA by commodity, and have estimated import and export demand models for the same 108 commodities.The results from the bounds testing approach to co-integration and ECM method indicate that in the long-run 41 export industries and 70 import industries are sensitive to the Australian RER.
In next, Bahmani-Oskooee, Shafiullah, and Islam (2017) criticised the study of Bahmani-Oskooee et al. (2005) for using linear model.Using a linear ARDL approach Bahmani-Oskooee et al. (2005) find that Australian bilateral trade, with each of the 23 partners, follows the J-curve effect in the model with the United Kingdom only.However, incorporating the ARDL model that allows for nonlinear adjustment of RER changes, they find that the J-curve effect is valid for four more partners (India, Italy, South Africa and United Kingdom).
Finally, Bahmani-Oskooee and Harvey (2019) have applied nonlinear models and asymmetric analysis approaches for testing the J-curve between Australia and her second largest trading partner-the United States-by using disaggregated data.However, this new approach also does not yield any significantly different outcomes to those of aggregate data or linear and symmetric analyses.They apply this new approach for the industry specific data to the trade flows of 123 industries traded between the United States and Australia and give evidence of an asymmetric J-curve in 28 industries.In addition, they find short-run asymmetric effects of RER changes on the TB of almost all studies, short-run impact asymmetric effects in 27 industries, and significant long-run asymmetric effects in 56 industries.
We would like to keep our literature review limited to Australian perspective.So, we have focused only the RER -TB nexus on the Australian economy.The findings from the above studies are mixed.All the papers before Bahmani-Oskooee and Wang (2007aWang ( , 2007b) ) suffers from aggregation bias where a significant price elasticities with one trading partner could be more than offset by an insignificant price elasticity with another partner.Neither do they follow proper econometric techniques.Moreover, some studies are not concentrated on the MLC meaning that the Australian AFF sector suffers from major knowledge gap regarding the MLC.
We have found only two studies on the MLC for Australian TB.Since they are pursued by aggregate data, these studies may suffer from aggregation bias.None of the papers above has concentrated on the MLC of Australian AFF products.According to our knowledge this is the first attempt to investigate the validity of the MLC for Australian AFF products.Since AFF commodities have some special properties that are dissimilar to industrial products, AFF products' TB may exhibit different behaviour than those of industrial commodities.Furthermore, Australia has a vast land property, unlike the most of the countries in the world.AFF products need intensive use land input in the production process.So, the study has more importance for Australia than other countries involved in AFF trade since a land abundant country has a relatively higher potential for AFF trade.Likewise, AFF products have contributed only 20% to total Australian exports and 10% in overall imports in 2020.Since there are huge unemployed land in Australia the country has much scope for further improvement in AFF TB by producing more AFF products.It is therefore clear that the study may have ample importance for the overall Australian economy and the agricultural sector in particular.

| THE MODEL, ECONOMETRIC TECHNIQUE AND DATA
Following Bahmani-Oskooee and Hosny (2013), Australian import demand from the partner country for commodity i would be as follows: where IM i is the imports of commodity i by Australia from the partner country, Y AUS is Australian real GDP, PIM i is the price of the imported product i and, PD AUS is the domestic price level in Australia.In this model, Australian real GDP and PIM i /PD AUS -the relative import price index-are assumed as the key determinants of imports.Considering the usual notion of the conventional economic theories, the sign of the estimated b and c should be positive and negative, respectively.
To obtain comparatively more stable estimated coefficients, Bahmani-Oskooee and Hosny (2013) also recommends to convert Equation (1) into a dynamic adjustment model by incorporating the short-run dynamic adjustment mechanism.Econometricians usually respecify Equation (1) by converting it into an ECM proposed by Pesaran, Shin, and Smith (2001).Thus, our dynamic specification stands as Equation (2) below, keeping coherence with the suggestion of Pesaran et al. (2001) and Bahmani-Oskooee and Hosny (2013).Our empirical estimation will be based on time series analysis using data for the period of 1988Q1-2020Q4 where the linear ARDL estimation technique is employed.
In Equation ( 2) Pesaran et al. (2001) recommends applying the standard F-test with new critical values to establish the joint significance of the lagged level variable as a sign of cointegration.They also tabulated new critical values to interpret the degree of integration of the variables used in the model (2).Undeniably, variables could be I(0) or I(1), which are very common features of almost all macroeconomic variables.Hence, we believe that there is no need for pre-unit root testing.However, we have calculated and found they are stationary either in level or in first-differenced form, and (ready to share them on request).Once Equation ( 2) is estimated, the coefficient estimates of the first-differenced variables reflect short-run effects.The long-run effects, that is, the income and relative import price elasticities in Equation ( 2) are obtained by the estimates of σ 1 and σ 2 that are normalized on σ 0. Next, we formulate the demand function of partner countries X (= USA, Japan, China, South Korea and Thailand) for Australian AFF commodity exports of i (EX i ) as a function of the country's X income (Y X ) as in (3): Again, we expect an estimate of b 0 and c 0 should be positive and negative, respectively.Furthermore, the ECM model associated with Equation (3) yields the following shape: Once again, Equation ( 4) is estimated, the short-run effects inferred by the estimates of coefficients related to first-differenced variables, and long-run effects are accompanied by the estimates of σ 0 1 and σ 0 2 normalized on σ 0 0 .The ML condition will be satisfied if the both conditions are satisfied which are normalized relative price elasticities, σ 2 and σ 0 2 are (i) negative and significant respectively in both model (2) and (4), and (ii) the summation of absolute values is more than one.The data, variables and traded AFF commodities are noted in Appendices A and B, respectively.

| RESULTS, DISCUSSION AND RESULTS SUMMARY
Now we can discuss the results of the estimated ECMs (2) and ( 4) for each of the five countries for individual AFF products that have been traded between Australia and five individual countries for the quarterly data over the period of 1988Q1-2020Q4.For this purpose we rely on the model of Bahmani-Oskooee and Hosny (2013) to estimate export and import demand elasticities, enforcing a maximum of four lags for each individual model for each first-differenced variable using the latest version of E-views.In this case, we have used the Akaike information criterion to define the optimum number of lags.Thus, every reported result is considered an optimum model.For our purpose, to infer the validity of MLC we need only long-run elasticities of the ECM model ( 2) and ( 4).So, we only report the estimated long-run coefficients of export and import demand functions.Since the short-run coefficients, are not related to our present concern; they are presentable on request.As estimated models are the same for each pair of countries, here we briefly point out the inference techniques.Since the model is an ARDL type of ECM, we first have to confirm a significant F-statistic about the co-integration among dependent (export or imports) and independent variables (income and relative price of exports and imports).This F-test statistic distribution table is tabulated by Pesaran et al. (2001) where a significance of calculated F-statistic indicates that relationship among the dependent and independent variables are meaningful (Table 2).In our estimations F-statistic shows significance overwhelmingly except one commodity (Z) and exports of three commodities (N, T and P1) by Australia to South Korea and Thailand, respectively, in case of imports of two commodities (T and U) from Thailand only.After achieving a confirmed co-integrated relationship we can set out for other diagnostic tests.We report here five other relevant diagnostic tests.We estimated the error component term (ECT) widely known as ECM tÀ1 1 with imposing optimum lags.It is known as speed of convergence to the long-run equilibrium where higher value of significant negative ECM confirms higher speed towards convergence.We have asterisked the commodity names mentioned in the first column of Tables D1-D5 in Appendix D when estimated ECM tÀ1 is found significant.The Lagrange multiplier (LM) test result is used for testing the presence of serial correlation.Since our quarterly data size is 132 and Breush-Godfrey serial correlation LM test statistic is distributed as ƛ 2 with 2 (two) degrees of freedom with a critical value of 5.99.The RESET test developed by Ramsey is also reported to infer the functional specification of the model.This statistic is also distributed as ƛ 2 with degrees of freedom 1 (one) for each model.We have also applied the CUSUM and CUSUMSQ tests to determine stability of the short-run and long-run coefficient estimates where stable coefficients are denoted by "S" and the unstable ones by "US."Finally, we also tested the goodness of fit of the every model.To test this, instead of R 2 , we have reported the numerical value of the adjusted R 2 .Since adding the unnecessary variable also increases R 2 , we believe that adjusted R 2 gives the true picture of goodness of fit of a model.The discussion of the results of the application of this econometric techniques for the five major AFF trading partner countries of Australia is noted below.
To this end, Table D1 of Appendix D reports the results of the imports and exports demand models of Australia and US bilateral commodity level AFF trade.It is notable that as our sample size is 132, t-statistic at least 1.646 (10% degrees of freedom) is considered as significant.By inspecting Table D1, it can be seen that the relative price of import coefficients has a negative and significant coefficient for 59.76% of AFF products imported to Australia from the USA, while in 56% of AFF products the income elasticity is positive, and the remaining 44% of them is negative.The negative elasticities imply that as the Australian economy grows, she produces very close substitute goods those belong to these 46% AFF products, which helps the country to lessen imports.However, net imports of Australian AFF trade will be increased as income increases since positive income elasticities are higher than negative ones.Lower part of the Table D1 presents the estimates of the demand by the US economy for Australian exports.It seems that as one of the major AFF trade partners of Australia, US income as a main long-run determinant of Australian exports in most of the AFF products, since 82% of total AFF trade carries a positive and significant coefficient.From Table D1 it is observed that 59.76% and 61.63% of Australian AFF imports and exports with the USA satisfy the MLC, that is, the sum of the absolute values of import and export price elasticities are greater than one, and relative price elasticities are individually negative and significant for both export and import demand functions.
Australian imports and exports with Japan are reported in Table D2 of Appendix D. The relative price of import coefficients has negative and significant coefficient for 17 products amounting to 62.77% of AFF products imports of Australia from Japan.Eleven of these products, amounting to 47% of products, have price elasticity with more than one meaning that they are relatively price elastic.Totally 91% AFF products imports have the income elasticity positive meaning that Australia has no substitute for those Japanese AFF products.That is, Australia has sufficient scope to focus on AFF trade to reduce imports from Japan to improve the TB.Now, lower part of Table D2 for the estimates of the Australian exports demand by Japanese economy can be considered.It seems that as one of the Australian major partners, Japanese income is a main long-run determinant of Australian exports in most the AFF products, since they carry positive and significant coefficients.Negative relative exports price elasticity with significant t-statistics are revealed by about 54% AFF exports by Australia to Japan.From Table D2 it is observed that about 62.77% and 53.82% of Australian AFF imports and exports, respectively, satisfy the MLC.
Table D3 of Appendix D reports the results of the fitted imports and exports demand models for Australian AFF products with China.By the visual inspection of import panel Table D3 it is clear that the relative price of import coefficients are negative and significant for 74.05% of AFF products imports of Australia from China.Income elasticity is positive, and significant for 100% AFF imports meaning that imports are increased as income increases of Australia as she has no substitutable capacity for these AFF products.In addition, the results of the bound tests confirm that Australian imports are co-integrated with Australian income or relative import prices or both for 100% of Australian AFF imports from China.Next, lower panel of Table D3 describes the AFF exports functions to China.Our estimation shows that for 88.99% AFF exports of Australia, Chinese income is a long-run determinant of Australian AFF products exports, since it carries a positive and significant coefficient.However, our main concern is regarding coefficients of relative exports price.It is significant and negative for more than 96% of exports to China which means that Australian AFF exports to China is highly price elastic.Finally, from Table D3 it is observed that for AFF 74% imports and 69% exports of Australia with China and Australia tends to satisfy the MLC.
Our next focus is on Table D4 of Appendix D to understand the status of Australian bilateral trade of AFF products with South Korea.Import panel of Table D4 shows that the relative price of import coefficients are negative and significant for about 53% of AFF products.While in 81% AFF products, the income elasticity is positive, meaning that Australian imports from South Korea increases as the Australian income increases.Positive and significant elasticities imply that the Australian economy has no direct or indirect substitutability for 81% imported products from South Korea.Next, the lower part of Table D4 review the Australian AFF products demanded by the South Korean economy.The relative price term carries a negative and significant coefficients for about 65% of total Australian AFF exports to South Korea which are also passed by the bound tests.From Table D4 it is observed that about 53% of Australian AFF imports and 65% Australian AFF exports satisfy the MLC.
Finally, we look at the Table D5 which does the same work for the AFF commodity level imports and exports demand models of Australia with Thailand.It is observed that the relative price of import coefficients have negative and significant coefficients for 52.29% of AFF products are imported to Australia from Thailand.For 82% AFF products the income elasticity is positive.The positive elasticities imply that as the Australian economy grows, the country needs to import more goods from Thailand as she is unable to produce substitutes for those AFF products, which would worsen her TB.To analyse AFF exports to Thailand, we look at the export related panel in lower area of Table D5.Like the above four countries Thai income is also a dominant long-run determinant of Australian exports for most of the AFF products, since it carries a positive and significant coefficient.In addition, 78.26% of AFF trade carries negative and significant relative export price elasticity, implying that Australian AFF exports to Thailand are highly price elastic.Among them, 78.12% of exports satisfy almost all econometric tests, indicating that the estimated coefficients of those functions are reliable.From Table D5, it is observed that for 52.29% and 78.12% Australian AFF imports and exports, respectively, with Thailand satisfying the MLC.The summary of the above discussed results is presented in Table 3 below so that decision on MLC can be done easily.
Table 3 shows what percentage of Australian AFF trade with her major five partners satisfy the MLC condition.It is clear that the majority of the AFF trade share confirms the efficacy of this condition so it can be claimed that the depreciation of the AUD either by market forces or by any other means would improve the Australian AFF TB in the long-run.This finding has important implications for policy makers of Australia.Moreover, the results suggest that some relevant policies, like export promotion for raising the return on exports and imports substitution measures to lower the import expenditure, must be taken into consideration to improve the AFF TB, and also ER depreciation through monetary and fiscal policy can be an easy way in this regard.

| CONCLUSION AND POLICY IMPLICATIONS
Economists have been discussing the issue of ER impact on TB at least for the last five decades.Hence, there is a great deal of research on this topic.One of the branches of this topic is investigation of Validity of MLC.MLC states that if the sum of the price elasticities of import and export demands of a country adds up to more than one, currency depreciation is expected to improve the country's TB in the long-run.The findings of this research state that this condition is valid for the majority portions of imports and exports by Australian AFF goods for each of her major five partners.It can be now concluded that depreciation of the AUD either due to market forces or for anything else that may improve the AFF TB of Australia with the major trading partners.
To avoid aggregation bias we estimated our proposed model by commodity-wise data.Furthermore, since the data used in this paper is time series in nature we have conducted unit root test to select the right model for estimation.Accordingly, we adopted the ARDL technique for the purpose of estimation.The results of our model indicate that the MLC phenomenon works for the fitted data and model; that is after passing through of depreciated RER, TB will ameliorated.This happens, perhaps, due to the increase and decrease of the profit margin of Australian AFF exporters and importers, respectively.Our estimation technique and results are reliable.Therefore, we believe any policy based on findings of this paper will bring good fortune for the Australian AFF TB.Furthermore, besides ER, income also plays an important role in determining the Australian AFF TB with the major five AFF trade partners.
Australia follows a principle of free market and liberalized borders in trade policy.The country also follows a market based flexible ER policy.Therefore, the country does not have any option to manipulate its ER for the improvement of trading account (TB), CA, foreign exchange reserve and so forth.In this backdrop, the country can use monetary and fiscal policy to convey the ER to its intended direction.In this way, Australian central bank has the scope to depreciate ER.Our findings suggest if the Australian central bank and government initiate proper monetary and fiscal measures, respectively, the AFF TB of Australia will be improved.
Last but not least, our findings oppose the proposition of Burda and Gerlach (1992) that durable products should be relatively more sensitive to the changes of ER than nondurable products.Since AFF products are mostly nondurable in nature and MLC is confirmed by major parts of exports and imports for each of the five largest AFF trade partners of Australia, it is clear that this proposition is not supported by our current study.

ACKNOWLEDGMENTS
Open access publishing facilitated by University of Southern Queensland, as part of the Wiley -University of Southern Queensland agreement via the Council of Australian University Librarians.[Correction added on 16 May 2022, after first online publication: CAUL funding statement has been added.]ENDNOTE 1 Short run parameters are estimated using the regular ECM as depicted is equation below The error correction model results indicate the speed of adjustment back to long run equilibria after a short run shock.The ECM integrates the short-run coefficient with the long-run coefficient without losing long-run information.Under ECM technique, the long run causality is depicted by the negative and significant value of the ECT coefficient α and the short run causality is shown by the significant value of other regressor variables.A same method is applied for export models too.LIST OF VARIABLES IMi = For each commodity i, IM is the volume of Australia imports from the trading partner country X.It is defined as the ratio of the value of Australian imports from the trading partner X over the respective import price index of commodity i. EXi = For each commodity i, EX is the volume of Australian exports to the trading partner country X.It is defined as the ratio of Australian exports to the trading partner country X over the respective exports price index of commodity i.
Bilateral imports and exports values are collected from the Department of Foreign Affairs and Trade of Australian Government.(Data period 1988Q1-2020Q4).
Y AUS = Australian real GDP.The data come from International Financial Statistics (IFS), published by IMF.
Y X = Real GDP of Australian trading partner country X (= US, Japan, China, South Korea, and Thailand).Data source: IFS.PMi = For each commodity i, PM is import price index of Australia, collected from IFS. PD = Domestic price level in Australia, CPI data is used as proxy data for PD collected from IFS. PXi = For each commodity i, PX is defined as export price index of Australia, collected from IFS.
P AUS = The price level in the US.CPI data used as proxy for P AUS collected from IFS.
Individual commodity price = This data is collected from Australian agriculture, water and the environment monthly publications.The supplied data are monthly.However, since these are flow data we take the quarter-end data only for our model.….
Note: *indicates ECM tÀ1 is significant with negative sign.
A P P END I X B: LIST OF AFF COMMODITIES TRADED BY AUSTRALIA CONSIDERED IN THIS RESEARCH A = Live animals B = Meat and edible meat offal C = Fish and crustaceans, molluscs, and other aquatic, invertebrates D = Diary produce E = Animal originated products F = Trees and other plants, live G = Fresh vegetables and certain roots and tubers H = Fruit and nuts, edible I = coffee, tea, mate, and spices J = Cereals K = Products of milling Industry L = Oil seeds and oleaginous fruits M = Lac; gums, resins and other vegetable saps and extracts N = Vegetable plaiting materials O = Animal or vegetable fats and oils and their cleavage products P = Cocoa and cocoa preparations Q = Preparations of cereals, flour, starch or milk R = Preparations of vegetables, fruit, nuts, or other parts of plants S = Miscellaneous edible preparations T = Beverages, spirits and vinegars U = Food industries, residuals, and wastes thereof V = Tobacco and manufactured tobacco substitutes W = Salt X = Milk, milk powder, butter, cheese etc. Y = Root crops and plants Z = Dry edible nuts and vegetables A1 = Round wood, swan wood, timber etc. B1 = Aquaculture, hatcheries and nurseries products C1 = Fertilizers originated from natural products D1 = Tanned, processed and raw hides E1 = Non-fish sea extracts F1 = Organic fertilizer and active agents G1 = Residues and waste from the food industries; prepared animal fodder I1 = Products of animal origin, not elsewhere specified or included J1 = Wool, fine or coarse animal hair, and animal hair yarn L1 = Fur skins, leather etc. M1 = Feathers and downs prepared N1 = Furniture O1 = Miscellaneous edible products P1 = Sugars and sugar confectionary L E D 4The results of the fitted model to Australian imports and exports with South Korea A PP E ND IX C: LITERATURE ON NEXUS BETWEEN EXCHANGE RATE AND TRADE BALANCE OF AUSTRALIA The results of the fitted model to Australian imports and exports with the USA The results of the fitted model to Australian imports and exports with Japan The results of the fitted model to Australian imports and exports with China A PP E ND IX D : DETERMINING MARSHALL-LERNER CONDITIONT A B L E D 1 … T A B L E D 3 The results of the fitted model to Australian imports and exports with Thailand T A B L E D 5