WELFARE EFFECTS OF PROTECTION AND ECONOMIES OF SCALE-THE CASE OF THE AUSTRALIAN AUTOMOTIVE INDUSTRY by MoonJoong Tcha and Takashi Kuriyama DISCUSSION PAPER 02.11 DEPARTMENT OF ECONOMICS THE UNIVERSITY OF WESTERN AUSTRALIA 35 STIRLING HIGHWAY CRAWLEY, WA 6009 AUSTRALIA Welfare Effects of Protection and Economies of Scale - The Case of The Australian Automotive Industry MoonJoong Tcha* and Takashi Kuriyama The existence of economies of scale requires different interpretations of the welfare effect of protection on the industry from conventional analyses. This paper finds that the Australian automotive industry has economies of scale, and carries out relevant welfare analysis of tariff. Using the fully modified Phillips-Hansen method, the paper estimates long-run elasticities and changes in consumer and producer surplus by tariffs. With the presence of economies of scale, it is reported that the net deadweight effect of tariffs is relatively small, while the redistribution effects of tariffs are substantial. Also, it is found that the argument that tariffs protect domestic employment is not plausible in the Australian automotive case, due to the existence of economies of scale. Key Words: Australian automobile indust1y, protection, economies of scale, welfare effects JEL Classification: F 12, F 13, F 14 * Main correspondence: Department of Economics, The University of Western Australia, Crawley, WA 6009, Australia. (email) mtcha@ecel.uwa.edu.au. The authors appreciate Ken Clements and Paul Miller for their constructive comments on the earlier version of this paper. Assistance from Patricia Wang and editorial assistance from K. Andrew Semmens is gratefully aclmowledged. Introduction Despite the global trend towards free trade, a few favored sectors in each country have shielded themselves from import competition. In Australia, since the first successful Australian car was built in 1901, the automotive industry has been regarded as important for the economy (Stubbs, 1972). Accordingly, the Passenger Motor Vehicle (PMV) industry has been the most heavily protected sector, together with Textile, Clothes and Footwear (TCF), by the Federal Government. While trade barriers for the industry have been reduced, partly due to the globalisation of the world automotive market from the late 1980s, the rate of tariffs on the PMV industry still remained relatively high (15%) in 2001. This study aims to analyze the welfare cost incurred by tariff imposition in the automotive industry for the past two decades. While the PMV industry has been heavily protected, in Australia, the trade deficit in the industry, has been increasing over time. The deficit reached $11.7 billion in 1999, which is a 76% increase from the deficit level in 1990 as shown in table 1. Although exports have been growing consistently over time thanks to the Export Facilitation Scheme (EFS), an increase in the demand for small cars and a reduction in the tariff rate are believed in to result in a more than proportionate increase in imports. Table 2 shows the local and import split of new PMV' s by sectors. In the small car sector, the share of imported cars, which was 36.6% in 1990, has grown dramatically reaching 92. 7% in 1999. This is explained by the withdrawal of local manufacturers such as Holden, Ford, Mitsubishi, Nissan and lately Toyota from small car production and the shift to the production of medium cars where they are more competitive (Capling and Galligan, 1992). No company produced small cars in Australia after Toyota's withdrawal in 1999. The small car sector has been dominated by Japanese manufacturers (with about 30% market share in 1999) as well as South Korean manufacturers. While locally produced vehicles dominated the medium car sector, Japanese car manufacturers still played an important role as Toyota and Mitsubishi operated locally. Table 2 also shows that the share of imported vehicles in the medium car market has been decreasing as domestic manufacturers have specialized in the production of medium sized cars. The crucial difference between the small and medium car sectors is that the total number of vehicles sold has been increasing in the small car sector together with the number of vehicles imported. On the other hand, total sales in the medium car sector have been 2 relatively stable but the proportion of imported vehicles sold. Imported cars outnumber domestically produced cars in the luxury car sector (this includes large cars and sports cars) as purchasers in this class preferred differentiated imported models. Protection and Welfare Cost The effects of protection of the automotive industry have been analysed by many researchers. Takacs (1994) assessed the net impact of the protection regime in the Philippine's motor vehicle industry and Okamoto and Sjoholm (2000) examined the productivity of the Indonesian automotive industry under high levels of tariff protection; however, they did not investigate and measure the impact of tariff changes on national welfare. The welfare effect on consumers and producers of tariffs or voluntary export restraints was estimated or simulated (for example, Goto (1992) and Hafbauer and Elliot (1994) for the U.S. automotive industry) and the implications of tariff structure for the motor vehicle industry was also explored (for example, Van Zyl and Kotze (1994) for the South African automotive industry). However, research on the welfare cost of protection of the automotive sector has rarely been carried out in Australia. For Australia, a general examination of the welfare cost of protection or trade policy was conducted (for example, Corden (1997) and Snape (1997)), and the welfare effect of tariff reform and the removal of monopoly was analyzed for some other industries (for example, Simmons and Smith (1994) for the sugar industry). Dixon (1978) developed theoretical concepts on how to measure the economic impact of tariffs (or the removal of tariffs) and Chand (1999) assessed the relationship between industry assistance and economic efficiency for various industries. Nonetheless, bearing in mind that the Federal Government of Australia regarded the PMV industry as crucial to the welfare of the economy and protected it intensively, it is surprising that a rigorous examination of the welfare cost of this protection has rarely been carried out. This paper analyzes the impact of trade barriers on the automotive industry on Australia's welfare by using a partial equilibrium model. The partial equilibrium model is used, as it enables relatively less complex and more transparent policy analysis by focusing on a few variables such as price and income (Francois & Reinert, 1997). Hafbauer and Elliot (1994) also used the partial equilibrium approach, arguing that it is simpler as the data requirements are far more modest 3 than those required for general equilibrium modelling, while still maintaining transparency. Completeness and complexity of the analyses might be lost by the partial equilibrium approach, however, the loss is not considered to be significant as the automotive sector does not explain a large portion of the Australian economy, regardless of the government's emphasis. While a variety of protection measures has been adopted in Australia over the last two decades, such as import quotas and local content requirements, tariffs have been the most prevailing policy tool used by the Federal Government. Therefore, this paper focuses on and measures the effect of tariffs on Australia's welfare. The demand for motor vehicles consists of two sources: new and second-hand motor vehicles. Australia has a very active second-hand car market. Maxcy and Silverston (1959) described the market for cars as 'a set of interrelated markets for a series of close substitutes'. As this study concentrates on the new car market only, changes in consumer's welfare due to fluctuations in the second-hand market prices are not investigated, although they are affected by the new car prices. It is widely accepted that tariffs have two effects: static and dynamic. Furthermore, static effects can be divided into direct and indirect effects (Productivity Commission, 2000). While direct effect refers to the change in consumer and producer surplus due to tariffs, indirect static effect, derived from direct effect, refers to the impact on job opportunities. That is, the higher domestic price caused by the imposition of tariffs will decrease the profitability of the consumer and user industry, and as a result decrease investment in that industry. Lower levels of investment will in turn adversely affect employment. On the other hand, the protected industry will enjoy greater profits and might spend more on investment, which will create more job opportunities. Dynamic effect includes a firm's opportunity to gain competitiveness in the world market or to accelerate the specialization facing larger markets over time. Dynamic effects are not easily observed and fall outside the scope of this study. Direct static effects, in particular consumer and producer surplus, will be computed in this paper, with some discussion on the indirect effects. Modelling The Automotive Industry and Data It is necessary to measure relevant elasticities of both demand and supply to estimate the direct welfare cost of tariffs using the conventional method. The Automotive Industry Council (1988) 4 claimed that the main factors influencing demand are vehicle prices, income and interest rates. However, many studies of demand analysis identified income and price as the most important determinants of demand. A few studies have concentrated on the supply equation for motor vehicles because there are a number of factors that affect the production of vehicles such as wages, input prices, technology and so on. Nevertheless, in this paper, we regard vehicle price to be the most influential factor, and hence, it is chosen as a single variable to avoid unnecessary complication. This study assumes that demand and supply curves have constant elasticities since isoelastic demand curves are convenient to work with (Pindick and Rubinfeld, 1997). More specifically, hyperbolic demand and supply curves, which are one of the most commonly used constant-elasticity-non-linear curves, are used: Demand: NR= a1x(PC)b'x(Rl)b2 Supply: CA= a2x(PC) 1 (1) 0 (2) where NR: the number of new car registrations, PC: the weighted average of the real price of cars, Rl: real national income, and CA: the number of cars produced in Australia. From equations (1) and (2), the price elasticity of demand is b1, the income elasticity of demand is b 2 and the price elasticity of supply is c 1, which are the coefficients for the variables reported from log-linear regression of the equation as + b1 In (PC1) + b 2 In (Rlt) + E Demand: In (NR1) = a1 Supply: In (CA,)= a2 + c1 In (PC 1) + v, (3) (4) Quarterly data was used in this study for 1984 to 1999. Data for CA was obtained from the Australian Bureau of Statistics (ABS) catalogues number 8363.0 and number 8301.0. NR and Rl data were taken directly from the ABS database, which are seasonally adjusted. ABS adopts SEASABS (Seasonal analysis to ABS standards) method, which is based on the X-11 ARlMA 5 package from Statistics Canada, to eliminate seasonal data problems (ABS, 2001). Real national income is calculated using the consumer price index (where 1989-1990 is the standard year). It is difficult to decide one representative price for automobiles, since automobiles are differentiated to a high degree. Even the same model has different prices depending on various options and characteristics. This study first selects models of automobiles that are frequently sold in Australia. Where some popular models are replaced with new models of vehicles, such as Hyundai Accent for Excel, the price of new model is used as the characteristics of those models are almost identical. Reflecting market shares, 15 models of small cars, 12 models of medium cars, and 19 models oflarge and luxury cars are selected 1• After the average price of each model (standard) is obtained (which are, as of 2001, $16,682 for small, $26,776 for medium, $44,444 for large), the average real price of cars for each year is calculated using the market share of each model, and the consumer price index. Estimating Demand and Supply Stationarity Tests for Variables Variables are called non-stationary or unit-root variables iftheir mean and variance change over time. Regression using non-stationary variables produces spurious results. Therefore, the unit root test ofrelevant variables is undertaken before proceeding with the regression analysis.2 The Augmented Dickey Fuller test (ADF) is adopted to test stationarity (Dickey and Fuller, 1981). ADF regression takes the form of k LiY, = $Yt-I +a +~t + :E 9j ilYt-j +Et, J=l where LiY 1 = Y1 - Y1_1, t is time trend, j is lag length, and s 1 is a sequence of independently and identically distributed random variables. The null hypothesis of the test is Ho: $ = 0. If the null hypothesis cannot be rejected, variable Y is non-stationary. The test results are summarized in table 3. Variables NR, CA and RI in logarithm (noted as LNR, LCA and LRl) are non- 1 The full list of models included in this study is available on request from the authors. We accept that the data used in this study, with 64 quarterly observations, are not sufficient to conduct unit root tests, however, believe that they would provide more reliable results. 2 6 stationary, while PC in logarithm ( = LPC) is stationary. Further tests are carried out for the three variables that have a unit root by taking the first difference. Table 4 exhibits the results. The two tests show that, after taking logaritlun, all the variables used, NR, CA, Rl and PC, are integrated of order 1 (I(l)). This study applies cointegration to resolve the unit root problem. Some pairs of nonstationary variables are said to be cointegrated if they wander closely due to the disequilibrium forces (Kennedy, 1992), which means that variables are non-stationary individually, but some linear combination of them is stationary. Error Correction Model of Automotive Demand and Supply As the variables with a unit root are identified, the order of the vector autoregressive (VAR) model needs to be determined. Firstly the estimation of demand curve is run using the logaritlun. Both Akaike Information Criterion (AIC) and Schwarz Bayesian Criterion (SBC) indicate that the order of VAR model should be four (table 5). The next step is to choose the number of cointegrating vector, r. Assuming that cointegration has unrestricted intercepts and unrestricted trends in the VAR, this study carried out the cointegration log-likelihood ratio (LR) test. The tests based on both the maximal eigenvalue of the stochastic matrix and the trace of the stochastic matrix reject the null hypothesis Ho: r = 0 but cannot reject Ho: r :> 1. Therefore, there is only one cointegration relationship between LNR and LRl. The demand relationship estimated by error correction model (ECM) is 3 3 dLNR, = a10 +all t + L);dLNR,_; + LY;dLRl,_; + 15, LPC, + $1EC (-1), i=I (5) i=l where EC is the error correction term. The results are reported in table 6. The coefficients for lagged variables dLNR t-i (= LNR,.; - LNR,.;.1, for i = 1,2, and 3) are all statistically significant and positive while those for lagged income variable dLRl t-i are insignificant. The direct relationship between the demand for motor vehicles and real income is not statistically proved. However, the error correction term, which includes the income variable (LRI) is statistically significant, which indicates that there exists a certain relationship between demand and income. The coefficients for price variables are also significant and negative, 7 implying that the change in prices adversely affects the demand for automobiles. This estimation reports conclusively that an increase in income and an increase in price respectively, affects demand for automobile positively and negatively, where the impact on demand disappears over time. The supply equation was estimated by using the same method. The order of VAR should be 1 based on the AIC and SBC. The number of cointegrating vectors is inconclusive: While the cointegration LR test based on both the maximal eigenvalue of the stochastic matrix and the trace of the stochastic matrix suggested the number of cointegrating vectors should be zero, the same test based on model selection criteria indicates that it should be one. This study presents the results from the both cases. When r is zero, as there is no integration term, ordinary least squared estimation is used for dependent variable dLCA on price variable LPC (as LCP is I(O)), that is, dLCA1 = a20 + 02 LPC1. (6)' The result is reported in table 7. The coefficient for price variable is positive, however, statistically insignificant. Table 8 reports ECM when the number of cointegrating vectors is 1, that is, dLCAt = U30 + Cl31t + 03 LPC1 + ~3EC (-1). (6)" In this case, the coefficient for price variables is reported as being negative, however, it is not statistically significant. The result from the ECM indicates that the quantity demanded is related to price and income significantly in the short-run. However, it fails to show the relationship between the quantity supplied locally and prices. Also, while the demand equation was estimated, the results are not easily applied to the welfare analysis. Therefore, this study alternatively estimates longrun relationships among relevant variables to explore whether welfare analysis is possible based on long-run elasticities. 8 Estimating Long-Run Elasticities Using Phillips-Hansen Method The fully modified Phillips-Hansen OLS model is particularly appropriate for estimation and inference when a single cointegration relation between a set ofl(I) variables exists (Pesaran and Pesaran, 1997). The variables used in this study are all I(l), except LPC which was found to be I(O). However, as discussed earlier, we have a limited number of observations, which might mislead the result of the unit root test. 3 Considering that the time series automobile price variable could be I(l), the fully modified Phillips-Hansen OLS method is applied to measure demand and supply elasticities. Parzen lag window is chosen where lag length is 1. As shown in table 9, price elasticity of demand is estimated as - 0.43 and income elasticity of demand as 1.27. These results are consistent with general observations of the relationship between the quantity demanded and prices and incomes. The figures are also consistent with other studies on automobile markets. For example, Hymans (1971) reported that for 20 years up to 1970, the elasticity of demand with respect to price was - 0.40 and that with respect to income was 1.00. However, the elasticity of supply with respect to price estimated by the Phillips-Hansen method is -0.38, which is substantially different from what might have been expected as it indicates that manufacturers produce more vehicles as the price (cost) of vehicles decreases. The result, that the Australian automobile industry has a long-run supply curve with negative slope, possibly indicates that existence of economies of scale is prevailing in the industry. Bloomfield (1978) explained that from an early stage in the history of the motor vehicle, economies of scale in both production and marketing have been significant in shaping the structure of the industry. The automotive industry in Australia has experienced dramatic restructuring since the mid-1980s, affected by government policy that aimed to improve the efficiency of the sector. Senator Button announced the requirement for the low volume producers in December 1986, to either stop production or to increase output (Snape, Gropp and Luttrell, 1998). As a result, the factories were automated and average costs decreased as more cars (but fewer models) were produced. These facts imply that it is highly probably that the Australian automotive industry could reduce average or marginal costs as it increased the number of automobiles produced. ' An investigation ofthe automobile price in Australia over time shows that it had increased with a clear trend until 1998, and then started to decrease. In addition, the test statistic for LPC is very close to the critical value of the ADF statistic as reported in table 3. These indicate that the automobile price is possibly 1(1 ). 9 Since the automotive industry is extremely capital intensive, an increase in investment on capital goods leads to the greater production efficiency. Empirical evidence indicates that investment expenditure for capacity increased from $53.4 million (1991-94) to $161.3 million (1995-98) as well as an increase in investment for efficiency from $292.8 million (1987-90) to $565.9 million (1991-94). It is doubtful however, whether the benefits derived from economies of scale is fully realized by this increase in investment. It was frequently observed that as the elasticity of supply with respect to price is negative, the tariff on automobiles actually reduced domestic production, contrary to the case of the decreasing returns to scale. Another valid argument for the downward supply curve can be found from the gradual reduction of tariff on imports of both completely assembled motor vehicles and motor vehicle parts. The abolition, in 1998, of local content requirements contributed to reducing the price of car parts. Lower prices of motor vehicle parts in turn facilitated the production of automobiles with lower costs, which contributed to shaping the long-run supply curve downward sloped. This effect might be more clearly explored once protection on related sectors, through tariff and nontariff protection measures, is considered as a whole. The welfare analysis of tariffs with the presence of economies of scale was intensively examined by Dixon (1978) in the context of general equilibrium. While traditional analyses reported relatively minor consumption and production effects of protection, Dixon (1978) pointed out that economies of scale, together with intra-industry specialization, would dramatically increase the effect. However, his measures of the production and consumption effects, arising from the reallocation of expenditures on goods and factors, are different from the welfare analysis conducted in this paper that concentrates on surplus in partial equilibrium models. Measuring Welfare Effects of Tariff - Discussion When a supply curve has a negative slope, the stability of the equilibrium is not necessarily guaranteed. In order to have a stable equilibrium, an increase in price should reduce excess d(Qo -Qs) 0 h demand, and vice versa. In other words, - - - - - < , w ere dP JO d(Qo -Qs) dP Tlo x Qo -ris x Qs p -0.43 X Q 0 +0.38 X QS "'---~~--~~ p where (5) rio= dQo ...!'..._ (the price elasticity of demand), and dP Qo dQs p T]s = - - - - (the price elasticity of supply). dP Qs Data used in this study shows that the number of locally produced automobiles (Qs) always failed to meet demand (Qo) throughout the period (1984-1999). Consequently, the value from (5) is always negative (i.e. d(Qo -Qs) < 0 ), which indicates that the Australian automobile dP market has a stable equilibrium and our welfare analysis is valid. Using elasticities estimated and data on quantity demanded, local production and supply and domestic and international prices, the net welfare effect of tariffs on automobiles can be calculated. This section will measure three kinds of welfare effects: changes in consumer surplus, changes in producer surplus and tariff revenue. Net welfare effects and deadweight Joss will be also calculated. Changes in consumer surplus over time are graphically shown as area [PAbcPw] in figure 1. Using the information obtained from regression and collected data, changes (Joss) in consumer surplus by tariffs in each year t can be obtained as which is, using the integration of the inverse demand curve for prices for each year t, 4.95 x f PA Pt -0.43x(RI,/27 dP, PW II where PA is 1he autarky price and Pw is the world (or free trade) price. As Australia is a small open economy, 1here would not be any terms of trade effect, and 1he world price and 1he domestic price have such relationship as PA = Pw + T where T are tariffs on automobiles, or txPw with t being the (ad valorem) tariff rate. The changes in consumer surplus measured are reported in the second column in table 10. As the elasticity of supply wi1h respect to price is found to be negative, the supply curve has a negative slope, which is steeper 1han the demand curve. Where economies of scale exist, an increase in domestic price due to tariffs reduces domestic production, which in turn decreases 'negative' producer surplus. This change in producer surplus is area [PAgePw] in figure 1. Area [PAgePw], which is the gain in producer surplus in each year t is computed as SrA [PA- Pw] + f, Sew SPA SdQ - [SrA- Srw] Pw, which is, using the integration of the inverse supply curve for prices for each year t, 13.0lx t P1 -o.is dP, and the results are reported in the third column in table 10. Government tariff revenue is the product of the total number of automobiles imported and the tariff on each automobile, which is (DrA-SrA)x(PA-Pw) as shown in figure 1 as area [gbdt] and reported in the fourth column in table 10. Area [get] in figure 1 is the negative deadweight loss due to the reduction of production. As area [gbdt] is total tariff collection, the net welfare loss is the difference between the deadweight loss of consumers and this deadweight gain (or negative deadweight loss) of producers, which is revealed in the last column in table 10. Total net deadweight loss for the 16 years (1984-1999) was calculated as $1.2 billion, which is on average $75 million each year and considered small (e.g. Hufbauer and Elliott, 19944). This 'small' amount of net deadweight loss is largely due to the 'deadweight gain' (or the negative deadweight loss) of producers. If the industry did not have economies of scale, the two deadweight losses should be added rather than 4 The size of the markets in Australia in this study and the US in Hufbauer and Elliott (1994) should be certainly considered in this comparison. 12 subtracted, and the total deadweight loss would be larger. As tariffs have been relatively low throughout the 1990s, this amount has been decreasing, reaching only $33 million in 1999. While tariffs prohibited the industry from benefiting from economies of scale, they also contributed to reducing negative producer surplus. In fact, consumer surplus has shrunk as much as $47 billion for the same period, which is on average about $3 billion per year. Therefore, if we consider consumer surplus only, this is a significant distortion. 5 The total consumer welfare loss due to tariffin 1999 was measured as $1.7 billion. This loss consists of two parts: area [PAbdPw} and area {bed]. The former area represents consumers' loss of surplus as they reduced their consumption of automobiles due to higher prices, which is about $1.1 billion. In 1999, as the number of new car registrations in Australia reached 552,575, each consumer who purchased a car paid about $2,000 ("' $1.1 billion/552,575) more than they would without tariffs. The latter area reflects consumers' loss by reducing their consumption of automobiles, which reached about $600 million. This $1.7 billion total loss (of consumers) in 1999, is equivalent to the total annual salary for 43,400 employees in the automobile and parts industry in Australia. In the same year, the total number of employees in that industry was only 51,694. The loss that consumers experienced is partly compensated for by the increase in producer surplus, which is about $2 billion per year. Taking into account that the net welfare loss due to protection is about $7 5 million per year, it should be pointed out that the protection policy for the automobile industry in Australia brought about more serious problems in terms of the redistribution of wealth than net efficiency loss. One of the arguments most frequently used to advocate protection policy is that tariffs can increase domestic employment. It is observed that the number of workers in the automobile industry in Australia has been decreasing since 1981, as the Federal Government demolished its protection policies. However, in order to make the argument valid that this decrease in employment is due to the removal of tariff protection, the industry supply curve should be positively sloped. If the industry has economies of scale as found in this study, a reduction in tariffs would decrease domestic prices, which in turn should increase employment. A close observation of data indicated that the total number of employees in the motor vehicle and related industries rapidly decreased until 1990, and then stagnated until 2000. Tariffs on passenger 5 As reported in table I, Australia had never exported automobiles more than$ 3 billion each year until 1999. 13 motor vehicles, components and replacement parts also rapidly decreased from 40% in 1990 to 15% in 2000. The total number of employees was not significantly affected by tariff-cuts. In addition, the motor vehicle manufacturing sector was the only sector which experienced a decrease of employment for the same period: other related sectors such as motor vehicle body manufacturing, automotive electrical and instrument manufacturing and automotive component manufacturing actually expanded or maintained existing levels of employment over the same period. Therefore, the reduction in employment in the motor vehicle industry should be regarded as the result of productivity increases (there has been an approximately 7% increase in labor productivity in this industry from 1992 to 1999) or capital-labor substitution, rather than the removal of tariffs. Conclusion The analysis of the welfare effects of protection policies for the Australian automobile industry resulted in some interesting findings. The industry' supply curve for 1984-1999 was found to be negatively sloped, which indicated the existence of economies of scale in the industry. When welfare effects of tariffs were measured through three parts as conventionally dealt with consumer surplus, producer surplus and government's tariff collection - it was reported that net welfare loss was extremely small. This can be expected once the industry demonstrates economies of scale, since the higher domestic price due to protection reduces negative surplus of producers, which in tum offsets part of the reduction in consumer surplus. Negative effects on consumers were in fact fairly large. In 1999, the loss of consumer surplus was equivalent to the total annual salary or wage for approximately 80% of workers then employed in the motor vehicle and parts industry in Australia. Also, there was no significant clue to conclude that the reduction of tariff progressed in Australia contributed to decreasing employment in the industry, at least for the last ten years. In fact, the reverse should be true: as the industry showed economies of scale, employment is supposed to increase as the domestic price falls due to liberalization. In summary, for the Australian economy, the efficiency loss due to tariff protection was relatively small. The distribution effect was however, much more serious. The amount of producers' gain and consumers' loss are both extremely high. 14 As the globalization of world automotive markets continues, the price of motor vehicles is expected to decrease as a result of active mergers and acquisitions in this industry and the trend of free trade. If the automobile industry in a country also has economies of scale, the combination of both lower prices and lower trade barriers will result in similar welfare effects as they produced in Australia. While the total net welfare gain is not as high as expected, there will be significant improvement in the distribution effect; sufficient to compensate consumers for their losses incurred as a result of protectionism. The exact magnitude will be, of course, dependent on the extent of each country's elasticities and the particular variables such as tariff rates and domestic automobile manufacturing costs. 15 References Australian Bureau of Statistics (2001). 5206.0 Australian National Accounts: National Income, Expenditure and Product. AGPS: Canberra. _ _ _ _ _ _ _ _ _ _ _ (2001). 5216.0 Australian National Accounts: Concepts, Sources and Methods, Appendix 3, Seasonally Adjusted and Trend Estimates. AGPS: Canberra. (2001). 6401.0 Consumer Price Index, Australia. AGPS: Canberra. _ _ _ _ _ _ _ _ _ _ _ (2001). 8301.0 Manufacturing Production Australia (October 1994- December 1996). AGPS: Canberra. _ _ _ _ _ _ _ _ _ _ _ (2001). 8363.0 Producer Bulletin No.7: Motor Vehicles, Motor Bodies and Trailed Vehicles Australia (January 1984 - December 1988). AGPS: Canberra. - - - - - - - - - - (2001). 8363.0 Production of Motor Vehicle (January1989 August 1989). AGPS: Canberra. _ _ _ _ _ _ _ _ _ _ _ (2001). 8363.0 Production of Transport Equipment, Australia (September 1989-August 1990). AGPS: Canberra. (2001). 8363.0 Manufacturing Production Australia Transport Equipment (September 1990- September 1994). AGPS: Canberra. (2001). 9303.0.55.001 New Motor Vehicle Registration (NMVR), Australia, Main Data. AGPS: Canberra. Bloomfield, G. (1978). The World Automotive Industrv. David & Charles Inc.: Vermont. Chand, S. (1999). "Trade Liberalization and Productivity Growth: Time-Series Evidence from Australian Manufacturing." Economic Record. vol.75, pp.28-36. Cordon, W. M. (1997). The Road to Reform: Essays on Australian Economic Policy. Addison Wesley Longman Australia: Melbourne. Department of the Prime Minister and Cabinet (1991). "Button Plan Stage 2-Building a Competitive Australia." AGPS: Canberra. Dickey, D.A. and W.A. Fuller (1981). "Likelihood Ratio Statistics for Autoregressive Time Series with a Unit Root." Econometrica. vol.55, pp.1057 - 72. Dixon, P. (1978). "Economies of Scale, Commodity Disaggregation and the Costs of Protection." Australian Economic Papers. vol.17, pp.63-80. 16 Goto, J. (1992). "Imperfect Competition and the Japan-US Automotive Trade." Ch.6 in M. Dagenais and P.A. Muet (eds), International Trade Modelling. Chapman & Hall: London, pp107-28 Hutbauer, G.C. and K.A. Elliott (1994). Measuring the Costs of Protection in the United States. Institute for International Economics: Washington D.C. Industry Science Resources (2000). Key Automotive Statistics. www.isr.gov.au _ _ _ _ _ _ _ _ _ _ (2001), Key Automotive Statistics. www.isr.gov.au Kennedy, P. (1992). A Guide to Econometrics. MIT Press: Cambridge, Massachusetts. Maxcy, G. and A. Silberston (1959). The Motor Industry, Allen and Unwin: London. Okamoto, Y. and F. Sjoholm (2000). "Productivity in the Indonesian Automotive Industry." ASEAN Economic Bulletin. vol.17, pp. 60-73 Phillips, P. and B. Hansen (1990). "Statistical Inference in Institutional Variables Regression with I(l) Process." Review of Economic Studies. pp 99-125. Pindick, R.S. and D.L. Rubinfeld (1997). Microeconomics. Prentice Hall: New Jersey Simmons, P. and Smith, A. (1994). "Producer Impact of Two Proposals for Reform in the Australian Sugar Industry." Economic Analysis and Policy. vol.24, pp. 57-72. Snape, R. (1997). "Tariffs, Then and Now: Lecture in Honour of the Late Bert Kelly." Australian Economic Review. vol.30, pp. 144-54. Snape, R., L. Gropp and T. Luttrell (1988). Australian Trade policy 1965-1997 -A Documentary History. Allen & Unwin: New South Wales. Stubbs, P. (1972). The Australian Motor Industry; A Study in Protection and Growth. Cheshire Publishing Pty Ltd.: Melbourne. Takacs, W. E. (1994). "Domestic Content and Compensatory Export Requirements: Protection of the Motor Vehicle Industry in the Philippines." World Bank Economic Review. vol.8, pp.127-49. Van Zyl, G. and Kotze, F. C. (1994). "An Analysis of the Tariff Structure of the Motor Vehicle and Related Industries in South Africa." Journal for Studies in Economics and Econometrics. vol.IS, pp.27-39. 17 Table 1. Australia's Balance of Trade for Automobiles Year Export (A$m) Import (A$m) Total deficit % change 1990 1,038.3 5,451.5 4,413.2 1991 1,162.2 5,023.2 3,861.0 -12.5 1992 1,248.7 6,398.8 5,150.1 33.4 1993 1,474.3 7,631.9 6,157.6 19.6 1994 1,537.6 8,881.1 7,343.5 19.3 1995 1,775.9 9,225.9 7,450.0 1.5 1996 2,263.5 10,690 8,426.5 13.1 1997 2,717.1 12,052 9,334.9 10.8 1998 2,574.3 14,494 11,919.7 27.7 1999 3,252.2 14,972 11,719.8 -1.7 Source: Ke~ Automotive Statistics 1999, JSR (2000) Table 2. New PMV Sales - Local and Import Split Small car Medium Car Large Car Local Import S.o.M.* 1990 95,883 55,390 36.6 209,314 53,562 20.4 14,179 35,089 71.2 76,949 64,402 45.6 178,664 22,785 11.3 7,900 38,222 82.9 1992 61,392 76,551 55.5 191,006 26,869 12.3 8,022 42,587 84.1 45,295 89,923 66.5 209,326 20,885 9.1 7,614 41,382 84.5 1994 36,320 109,186 75.0 235,991 21,542 8.4 9,275 48,384 83.9 1995 26,020 145,112 84.8 237,574 21,439 8.3 11,161 47,066 80.8 1996 24,452 158,925 86.7 230,765 20,066 8.0 9,729 48,121 83.2 1997 22,348 205,830 90.2 222,415 26,018 10.5 8,906 54,836 86.0 1998 21,889 229,500 91.3 243,705 24,900 9.3 7,767 56,599 87.9 16,870 212,849 92.7 239,266 7.4 9,716 54,751 84.9 Year 1991 1993 1999 Local Import S.o.M.* 19,123 • S.o.M. stands for "Share of Import" in percentage. Source: Ke~ Automotive Statistics 1999, !SR (2000) 18 Local Import S.o.M.* Table 3. Summary of ADF Test (in Logarithm) Test statistic* Variable ADF statistic** Unit root Order LNR -0.4233 -2.9006 Yes ;eI(O) LCA -2.1069 -2.9241 Yes ;tI(O) LRI -0.9693 -2.9006 Yes ;eI(O) LPC -3.0153 -2.9006 Yes =I(O) •Value for ADF (!) •• 95% critical value for ADF statistic Table 4. Summary of ADF Test (First difference, in Logarithm) Variables Test statistic* ADF statistic** Unit root Order DLNR -4.8482 -2.9012 No =I(l) DLCA -5.0893 -2.9256 No =I(l) DLRI -9.3701 -2.9012 No =I(l) "Value for ADF (I) ""95% critical value for ADF statistic Table 5. Choice Criteria for Selecting the Order of the VAR Model LL AIC SBC Adjusted LR test 4 515.9146 476.9146 431.7236 ---------- 3 442.7441 412.7441 377.9818 120.9752 [.000) 2 436.2800 418.2280 404.3230 131.6626 [.000] 1 430.2280 418.2280 404.3230 141.6686 [.000] 0 121.4224 118.4224 114.9462 652.2271 [.000] Order 19 Table 6. ECM for Demand Based on Cointegrating VAR (4): 1984-1999 (dependent variable= dLNR) T-Ratio [Prob] Coefficient Standard Error 7.3202 1.3706 5.4309 [.000] Trend .0064 .0012 5.1441 [.000] dLNRl .0321 .1140 .2826 [.000] dLRll -.0191 .1081 -.1766 [.860] dLNR2 .4961 .1335 3.7152 [.000] -.1148 .1230 -.9334 [.354] .5793 .1371 4.2243 [.000] dLR13 -.0643 .1049 -.6128 [.542] EC (-1) -.5250 .0976 -5.3805 [.000] LPC -.3453 .0780 -4.4246 [.000] Regress or Intercept dLR12 dLNR3 EC(-1) = 1.0000*LNR-0.042l*LR1 ·-------------------------------------------------------------------------------------------------------------------------R-Squared .4147 R-Bar-Squared .3337 S. E. of Regression .0555 F-Stat. F( 9, 65) 5.1171 [.000] Mean of Dependent Variable .0055 S. D. of Dependent Variable Residual Sum of Square .2002 Equation Log-Likelihood Akaike Information Criterion 105.7942 Schwarz Bayesian Criterion DW-Statistic System Log-Likelihood 1.9917 20 .0678 115.7942 94.2068 297.5374 Table 7. Ordinary Least Squares Estimation for Supply: 1984-1999 (dependent variable= dLCA) Regress or Intercept dLPC Coefficient Standard Error -.1380 .2288 -.6030 [.549] .0295 .0503 .5859 [.561] R-Squared .0070 R-Bar-Squared S. E. ofRegression .0801 F-Stat. F( 1, 49) Mean of Dependent Variable -.0041 Residual Sum of Square .3146 Akaike Information Criterion 55.3884 DW-statistic T-Ratio [Prob] -.0133 .3433 [.561] S. D. of Dependent Variable .0796 Equation Log-Likelihood 57.3884 Schwarz Bayesian Criterion 53.4566 2.4659 Table 8. ECM for Supply Based on Cointegrating VAR (1): 1984-1999 (dependent variable= dLCA) Regressor Coefficient Standard Error T-Ratio [Prob] 3.4482 1.4900 2.3142 [.025] -.8535E-3 .0021 -.4015 [.690] EC (-1) -.2096 .0758 -2. 7663 [.008] dLPC -.0178 .1445 -.1230 [.903] Intercept Trend EC(-1) = 1.4154*LCA R-Squared .1482 R-Bar-Squared S. E. of Regression .0758 F-Stat. F( 3, 47) .0938 2.7255 [.055] Mean of Dependent Variable -.0041 S. D. of Dependent Variable Residual Sum of Square .2698 Equation Log-Likelihood 61.3033 Schwarz Bayesian Criterion 53.4367 System Log-Likelihood 61.3003 Akaike Information Criterion DW-Statistic 57.3003 2.2077 21 .0796 Table 9. Fully Modified Phillips-Hansen Estimates for Demand and Supply: 19841999 Regressor Coefficient Standard Error T-Ratio [Prob] Demand (Dependent variable= LNR) Intercept 4.9481 .5456 9.0687 [.000] LPC -.4292 .0590 -7.2750 [.000] LRI 1.2731 .1058 12.0307 [.000] 13.0087 .3198 40.6762 [.000] -.3790 .0702 -5.3953 [.000] Supply (Dependent variable= LCA) Intercept LPC 22 Figure 1. Welfare Effect of Tariff with Economies of Scale p a h c Pw Spw DrA Dpw 23 D Q Table 10. Estimated Welfare Effects of Tariff ($million) Year Changes in Consumer Surplus Changes in Producer Surplus 1984 -3,851 3,057 693 -100 1985 -4,475 3,370 975 -130 1986 -3,734 2,983 656 -96 1987 -3,649 3,049 518 -83 1988 -3,487 2,782 630 -75 1989 -3,572 2,679 809 -84 1990 -3,442 2,352 999 -91 1991 -2,630 1,768 764 -67 1992 -2,693 1,713 909 -71 1993 -2,686 1,688 931 -67 1994 -2,868 1,742 1,056 -70 1995 -2,763 1,545 1,149 -69 1996 -2,489 1,333 1,096 -59 1997 -2,302 1,075 1,170 -56 1998 -2,111 983 1,081 -47 1999 -1,701 807 861 -33 Total -47,452 32,927 14,328 -1,197 Tariff Revenue Net Effect NOTE: Tariff rates between 1984 and 1987 are treated as being 57.5% although various policies such as tariff quotas and trigger tariffs were implemented. 24 DISCUSSION PAPERS Department of Economics The University of Western Australia 02-10 Clements, K. W. Three Facts About Marijuana Prices 02-09 Tcba,M. Pershin, V. Reconsidering Performance at the Summer Olympics and Revealed Comparative Advantage 02-08 Cbiswick, B. R. Lee, Y. L. Miller, P. W. Longitudinal Analysis of Immigrant Occupational Mobility: A Test of the Immigrant Assimilation Hypothesis 02-07 Chiswick, B. R. Lee, Y. L. Miller, P. W. Immigrants' Language Skills: The Australian Experience 02-06 Chiswick, B. R. Lee, Y. L. Miller, P. W. Family Matters: The Role of the Family in hnmigrants' Destination Language Acquisition. 02-05 Chiswick, B. R. Lee, Y. Miller, P. W. Immigrants' Language Skills and Visa Category. 02-04 Connolly, T. Butler, D. Searching for the "Regret" in "Regret Theory". 02-03 Groenewold, N. Hagger, A. Madden, J. R. The Efficiency of Federal Inter-Regional Transfers Under a Regime of Politically-Maximizing Regional Governments. 02-02 Groenewold, N. Tang, S. H. K. Wu, Y. The Dynamic Interrelationships Between the Greater China Share Markets. 02-01 Lan, Y. Aspects of Exchange-Rate Economics