1 Introduction Local content (LC) schemes have been used by various countries for many years. According to a UNIDO study 1 mainly developing countries

Transcription

1 Do Local Content Schemes Encourage Innovation? y Herbert Dawid Marc Reimann z Department of Management Science University of Vienna Abstract In this paper we study the eects of content protection on the technological progress in the protected industry. We use a very simple twoperiod model in which one nal good producer may buy his inputs from a global supplier who operates on the competitive world market and a local supplier operating as a domestic monopolist. The results from our simulation studies suggest that the eects of content protection mainly depend on the strength of the capital market, the protected rms' initial costs and the capital intensity of the protected industry. Having determined these eects we compare dierent value added local content schemes with each other and examine the impact of a phase-out of such a scheme on the incentives to invest. y We would like to thank Bernd Bullnheimer for his contributions to this research. Financial support from the Austrian Science Foundation under Grant F \Adaptive Information Systems and Modelling in Economics and Management Science" is gratefully acknowledged. z Corresponding author: Marc Reimann, Department of Management Science, University of Vienna, Brunner Strasse 72, A-1210 Vienna, Austria; fax: ; reimann@pom.bwl.univie.ac.at. 1

2 1 Introduction Local content (LC) schemes have been used by various countries for many years. According to a UNIDO study 1 mainly developing countries use this policy to gain more control over the operations of foreign owned companies without having to share the risk. Some other goals are the strengthening of an infant industry, technological progress, improvement of the labour force and the improvement of the balance of trade. Basically local content regulations require "a certain amount of inputs (by value or quantity) in a given industrial output to be of domestic origin". 2 Failure to comply with such a regulation is sanctioned by imposing "economic penalties, such as payment of a high tari rate on all intermediate imports." 3 Among the industries protected by LC schemes are vehicle assembly, automotive component production and consumer durables. These industries are highly capital intensive, i.e. it takes a long time until start-up costs are amortised, but they require lower skills than the production of cars itself and the barriers to entry, such as a minimum eective scale, are lower. These factors make them preferred targets for countries on their way to industrialisation. Some examples of LC requirements in developing countries can be seen in the following table 1. Country Sector LC requirements Argentina Automobiles 88% Brazil Automobiles % Egypt Automobiles 40-60% Mexico Domestic appliances 75-90% Republic of Korea Automobiles 20-95% Source : UNIDO(1986) Table 1: LC requirements in some developing countries A lot of the research on content protection has focused on the macroeconomic welfare eects of LC plans in dierent market settings. Grossman (1981) made one of the rst analytical contributions. He studies the eects of content protection under perfect competition as well as under monopoly. Mussa (1984) favourably compares content protection with taris. He also 1 UNIDO UNIDO 1986, p.2 3 UNIDO 1986, p.2 2

3 shows that gains from improvements in technical eciency will be biased towards such improvements that lead to a reduction of imported inputs. Vousden (1987) and Hollander (1987) consider monopolistic settings, while Krishna and Itoh (1988) study an oligopolistic input market and show that under certain conditions the domestic supplier earns lower prots under LC regulation. Richardson (1991) considers a duopsony in the nal good market in which a local content requirement is operating and shows the consequences for the domestic producer of this nal good. In all the papers mentioned above surprisingly little is said about the eects of protection schemes on the innovation incentives in the protected industries, although technological progress is widely considered as one main goal of such a policy. This is even more surprising if one considers the extensive literature on the eects of public market intervention - by means of price regulation - on innovation incentives. For example, the studies on price regulated monopolies include analytical work (Sweeney (1981)), empirical evidence (Lyon (1995)) and simulation models (Mayo,Flynn (1988)). The focus here is on the negative incentives of price regulation on innovation and how these incentives are weakened by the regulatory lag. In a slightly dierent setup, Teisberg (1993) considers a situation where due to the long lead times and uncertainty about future regulation the future return on current investment is unknown. Whereas this last setup shares some characteristics with the model we are interested in, there is of course a main dierence between all these models, dealing with price regulation, and the examination of markets under content protection. As stated above, to our best knowledge, no formal studies have addressed the eects of LC protection on innovation incentives. In this paper we intend to ll this gap and focus exactly on the question under which circumstances content protection encourages investments in process innovation. It has been argued in defense of content protection that as long as the local industry is small with weak nancial backing large investments in process improvements are only possible if the market is protected. In an unprotected environment the necessary investments could lead to bankruptcy even before they become eective 4. To formalize this argu- 4 e.g. the UNIDO study form 1986 argues that content proctection... is as much a qualitative as a quantitative transformation of the national capacity, not just to produce a product, but to master new processes and to enhance the country's capacity in such a way as not only to be able to reproduce a given production process, but to build on this technology and entrepreneurship and these skills to contribute to creating an ever more 3

4 ment we will in this paper consider an intertemporal model (for reasons of simplicity we will restrict ourselves to two periods) where the eects of investment innovation and technology improvements become eective only with a time lag. The framework we use is that of a manufacturer who may buy his inputs from a single local supplier or on the world market. Whereas the world market is in our simple model just represented by the resulting world market price, the inuence of dierent market modes on the investment incentives for the local supplier are our subject of study. In particular we compare the case where the local input market is protected by a LC requirement with the case of an unprotected market. The local rm cannot perfectly anticipate the market conditions { in particular the world market price { for the time before the investment pays o and therefore cannot exactly predict its prots. Thus, there is a potential danger of bankruptcy which is of course larger the larger the initial investment was. This generates two opposite investment incentives for the local supplier, a negative one stemming from the increase in the probability for bankruptcy and a positive one stemming from increased expected prots in the second period. The decision problem for the local supplier is to nd an investment that maximises his expected prots over both periods. Intuitively protection of a market should decrease both incentive eects but a priori it is hard to estimate under which conditions the overall eect of market protection is to encourage or discourage investment in process improvements. In what follows we will use this simple model to rigorously address this question. Our paper is organised as follows. In section 2 we present our market model in general before we specify the cases of the content protected and open market respectively. We report and discuss our simulation results in section 3 and conclude in section 4 with some nal remarks and possible extensions of the model. 2 The Model In our model we consider an input market with one local supplier operating in a domestic monopoly and a foreign supplier operating on the competitive world market. Both rms produce the same, homogenous good which is the single input into a nal good. This nal good is produced by a nal good producer situated in the country of the local supplier. complex and sophisticated industrial structure. 4

5 Dening x jl and x jf as the shares of the input supplied in period j by the local and the foreign supplier respectively we can write x jf + x jl = 1 j = 1; 2: (1) The nal good producer's demand for this input depends upon both supplier's prices. In our model the foreign supplier's price is a priori unknown to the local supplier. Thus, the local supplier treats it as a stochastic variable. Given his cost function c jl (x jl ) = j x jl ; (2) where j is a parameter for the eciency in period j, the local supplier charges the price p jl which maximises his expected prot IE( j ) = IE[(p jl? j )x jl (p jl ; p jf )]: (3) We denote the optimal price by p jl( j ; p jf ), where p jf is the foreign suppliers' price. In order to be able to analyze the intertemporal eects of investment we use a two-period model to study the investment behaviour of the local supplier. By investing an amount of in period 1 the local supplier can increase his eciency in period 2 in the following way: 2 = b 1 + (1? b) 1 e? : (4) Here is a parameter characterising the capital-intensity of the industry and b 1 is the highest reachable eciency. The local supplier has to make the decision of how much to invest at the beginning of period 1 without knowing his rst period prots. If the investment exceeds the sum of his prot in period 1 plus a credit limit granted to him he goes bankrupt in the rst period and does not survive until period 2. If he invests less than he can borrow, i.e. less than the credit limit, he faces no risk. However, if he invests more than that, the local supplier goes bankrupt whenever > D + M(p jl( j ; p jf )? j )x jl (p jl ; p jf); (5) where M is the demand for the nal product and D denotes the credit limit. Given the local suppliers' price and an investment exceeding the credit limit in period 1 it depends on the prot, which in turn depends on the sold amount 5

6 x 1L whether this inequality holds or not. Furthermore the amount x 1L depends on the world market price p 1F. As stated above this price is treated as a stochastic variable and the local supplier estimates the probability that he will survive period 1 with probability q( ) = 8 < : 1 D IPfM(p 1L( 1 ; p 1F )? 1 )x 1L (p 1L ; p 1F )??Dg > D: (6) Thus, while a high investment increases expected prots in period 2 it lowers expected prots in period 1 and the probability of surviving period 1. The problem for the local supplier is to nd a level of investment that maximises the expected prots over both periods. Obviously, the optimal level of investment depends on the shapes of the functions x jl (p jl ; p jf ) and p jl( j ; p jf ). The shapes of these functions are determined by the type of market regulations on the local market, in particular by the fact whether it is protected by LC schemes or not. The main goal of our study is to compare the optimal investment policy of the local supplier in two dierent settings. In the rst setting the domestic input market is protected by a content requirement while in the second case no government protection is in use. Also, we will analyse the eect of the use of several dierent LC schemes. 2.1 The Market with Local Content Protection We consider a value-added LC requirement stating that a certain ratio between the value of the local input and the total value of the nal good has to be met. Let denote this ratio then the requirement can be written as p jl x jl p jl x jl + p jf x jf j = 1; 2; (7) which can be reformulated to x jl x jl := p jf (1? )p jl + p jf : (8) At given input prices the producer of the nal good has to buy a minimum amount of x jl from the local supplier to satisfy the LC requirement. It is obvious that the producer will purchase all the inputs locally if p jl p jf. 6

7 From equation (8) one can see that the producer's demand for inputs from the domestic supplier falls if the price of these inputs increases. However this demand never actually becomes zero if there is no upper limit for the price the local supplier can charge. To make sure that the producer of the nal good adheres to the LC requirement, the government has to impose a penalty for failure to comply, such as a tari on all imported inputs. If the rm ignores the LC requirement, the cost of buying everything on the world market is p jf (1 + T ), where T denotes the tari rate. Adhering to the regulation causes costs of (p jf x jf + p jl x jl ). Using these two formulas and (8) we get the local supplier's upper price limit as ( pjf (1 + T )( )? T (1? ) > 0?T (1?) p jl = (9) 1? T (1? ) 0: If he charges more than this the manufacturer would prefer to violate the LC requirement and pay the tari. At this upper price limit the quantity supplied is x jl = max(0;? T (1? )): (10) This quantity depends on the LC scheme. It becomes zero if T > 1? : (11) The higher the LC requirement is, the higher the tari rate set by the government can be. However, if the tari rate is set too high, the local supplier supplies x jl = 0 and still receives a payment from the nal good producer who is better o by paying to the local supplier and thus "satisfying" the LC requirement than by paying the tari rate. 5 In the remainder of this paper we will consider only LC schemes that satisfy T. 1? The demand function for the local supplier's products can then be written as x jl (p jl ; p jf ) = 8 >< >: 0 p jl > p jl p jf p jf +(1?)p jl p jf p jl p jl 1 p jl < p jf : (12) 5 In such a case the nal good producer would have to pay p jf to the foreign supplier and p jf T to the local government. From (7) one can calculate the minimum amount Z which is necessary to comply with the LC scheme as Z = pjf. As this is less than p 1? jf T if T > it is advantageous for the nal good producer to make this payment to the 1? local supplier. 7

8 It can be easily seen that if the suppliers' price is above the world market price (i.e. if p jf < p jl ), and thus the quantity of sold goods is determined by the local content rule, the suppliers' prot is increasing in p jl as long as the price stays below p jl. Thus, the optimal decision for him is to charge the maximal price which guarantees positive sales, namely p jl. It is obvious from these arguments that the local supplier always charges p jl if the market price is below his costs (i.e. p jf j ), however if the supplier is ecient enough, i.e. p jf > j, he has to decide between charging the world market price p jf and charging the LC price p jl. In the rst case he supplies the nal good producer fully, while in the second case he supplies x jl =? T (1? ). The local suppliers' prot associated with the price p jf is (p jl ; p jf ) = M(p jf? j ): If the supplier charges p jl the associated prot is (p jl ; p jf ) = M(p jf (1 + T )? j [? T (1? )]): The decision which price to chose depends on the LC scheme and the market price. Straightforward calculations show that (p jl ; p jf ) (p jl ; p jf ) holds whenever T 1?. If T < 1? the prot associated with the price p jl is higher than the one associated with p jf whenever p jf < j (1? )(1 + T ) (1? )(1 + T )? T := j 1 : (13) Note that the expression for 1 goes to innity as T! 1?. Furthermore, given our assumption that (11) does not hold, a case where T 1? can only occur if > 0:5. Summarising, the optimal pricing policy of the local supplier in a protected market is given by p jl( j ; p jf ) = 8 < : max(p jl ; j ) p jf 1 1 max(p jf ; j ) p jf > 1 1 : The expected prot for each period can then be written as IE( j ) = IE h Mmax[0; (p jf? j ); (p jl? j )(? T (1? ))] i : (14) Having characterised the local supplier's optimal pricing policy we will now deal with the optimal investment decision in period 1. By investing an 8

9 amount of at the beginning of period 1 the rm can increase its eciency in period 2 to 2 as formulated in (4). The probability of surviving period 1 formulated in (6) depends of course on the market price p jf. To calculate the analytical expressions for the minimal market price level which guarantees survival in period 1, we have to distinguish between two cases: in one case this minimal level is in the interval where the supplier charges the market price, while in the second case it is in the interval with an optimal price p jl. Which of these two cases applies depends on whether the supplier survives with a market price of p jf = 1 1 where he is indierent between charging and charging p jl. From these considerations we get p jf q LC ( ) := 8 >< >: 1 case 1 IPfp 1F?D + 1 g case 2 ) 1+ T x 1L Mx 1L g case 3. Mx 1L (1+ T x 1L IPfp 1F 1 +?D (15) case 1: D case 2: > D; 1 (?D)[1?(1+T )] T M case 3: > D; 1 < (?D)[1?(1+T )] : T M Using (14) and (15) we can write the objective function for the supplier's decision problem as max IE( ) (16) = IE[Mmax[0; (p 1F? 1 ); (p 1L? 1 )(? T (1? ))]]? + q LC ( )IE[Mmax[0; (p 2F? 2 ); (p 2L? 2 )(? T (1? ))]]: 2.2 The Open Market In the open market the producer of the nal good buys the inputs from the supplier charging the lowest price. We assume that the nal good producer prefers the local rm, if the local and the foreign supplier charge the same price. The demand function for the local supplier's products can be written as x jl (p jl ; p jf ) = 8 < : 1 p jl p jf 0 p jl > p jf : (17) 9

10 It is easy to see that the prot maximising price for the local supplier induced by this demand function is p jf as long as p jf > j. The optimal price therefore reads p jl( j ; p jf ) = max(p jf ; j ). The local supplier's expected prot in each period can then be written as IE( j ) = IE[Mmax(0; p jf? j )]: (18) As in the case of the content protected market the supplier never faces a loss if he does not invest. If we extend the model to allow investment the probability of surviving the rst period is q OM ( ) := 8 < : 1 D IPfp 1F 1 +?D Mx 1L g > D: Using (18) and (19) we can write the objective function as (19) max IE( ) (20) = IE[Mmax(0; p 1F? 1 )]? + q OM ( )IE[Mmax(0; p 2F? 2 )]: 3 Simulation Results The objective function of the supplier in both the local content and the open market case is a rather intricate non-concave function of the level of investment. It is therefore not possible to give a closed form solution of the problems (16) and (20) and provide an analytical sensitivity analysis. Thus, we will have to rely on numerical studies in order to characterise the eects of dierent market forms and local content schemes on the optimal investment decision. Such numerical studies will be presented and discussed in this section. Our main focus will be on the comparison of the optimal investment in the open and the protected market respectively. We will also compare dierent LC schemes with each other. The results presented here are representative in the sense that as far as we can tell all eects are qualitatively robust with respect to variations of the actual parameter settings. For the runs reported below the following settings were used: the world market price in both periods follows the distribution p jf N(20; 4), demand for the nal good is 100 units (M = 100) and the maximum increase in eciency is 25%, i.e. b = 0:8. 10

11 3.1 Optimal Investment in Open and Protected Markets The simulation runs in this section were all performed with values of = 0:4 and T = 0:2. 6 To compare the investment behaviour of the local supplier in an open and a protected market under dierent constellations we singled out the initial eciency 1, the capital intensity of the industry and the credit limit D as factors that aect the decision to invest. The credit limit D has an impact on the probability to survive period 1. The more a rm can borrow on the capital market, the more it will be able to invest without going bankrupt. If the rm can borrow enough to nance its prot maximising investment it faces no risk of bankruptcy. The capital intensity of the industry determines the eect of an investment on the eciency in period 2. Higher capital intensity means that a higher investment is needed to achieve the same increase in eciency. The initial eciency 1 aects period 1 prots and thus the probability to survive the rst period. It also aects the eciency and thus the prot in period 2. An inecient rm will have less capital to invest as its prots in the rst period are lower. It also has to invest more to achieve high eciency in period 2. Again this lowers the probability to survive period 1. Following the argument lined out in the introduction, namely that especially for inecient local rms content protection is crucial to provide incentives to invest in process improvements, we will focus on the impact of the parameter 1 on the optimal investment. Of course the optimal investment level also depends on the risk of going bankrupt and therefore on the credit limit D. Thus, we rst investigate the impact of high respectively low credit limits on the optimal investment decisions before we go on to study the merits of market protection for dierent levels of initial eciency 1. As pointed out above the credit limit inuences the risk of going bankrupt and in general a high credit limit should increase the incentives to invest in cost reductions. How large this eect is in protected markets compared to open ones can be seen in gure 1. The initial cost of production of the rm is assumed to be low compared to the expected market price ( 1 = 14 in gure 1a) and 1 = 16 in 1b)) 6 In order to consider the least restrictive setting, we chose these values to satisfy the conditions T < 1? and T. Under this parameter constellation the local supplier 1? charges the price p jl only if (13) is satised. Otherwise he charges the world market price p jf. 11

12 a) 1 = 14 D b) 1 = 16 D Figure 1: Optimal investment in open and protected (bold line) markets for dierent values of the credit limit D. and the capital intensity parameter is set to = 0:02. Let us rst focus on gure 1a). We can see clearly that a variation of D in the interval [0; 100] does hardly aect the optimal investment decision in a market with content protection, but does so in an open market. This can be easily understood. In a market with content protection the optimal level of investment, if the danger of bankruptcy is neglected, is approximately 87:5 for this parameter constellation. This amount can be nanced by the rst periods' prots whenever the market price in the rst period is above approximately 10. Given the distribution of the market price, this happens with a probability of virtually 1. Thus, when determining the optimal investment the credit line is almost irrelevant. On the other hand, in case of the open market there is a positive probability for bankruptcy whenever > D. This is due to the fact that with such a market form the local supplier can at most charge the market price and thus prots tend to zero when the market price approaches 1. Thus, the decision whether a higher amount should be invested than is covered by the credit line is not trivial. The local supplier has to take into account two opposing eects. The cost reducing eect of an increase of to a level above D has to outweigh the expected costs associated with the risk of bankruptcy. We can see that for a small credit line indeed the rst eect is stronger whereas for values of D 76 the rm changes policy and avoids the risk of bankruptcy by setting = D up to the level = 86 which is the optimal investment level in an open market without any risk of bankruptcy. The optimal investment level here is higher in the content protected market 12

13 than in the open one even if survival is guaranteed. However, again two opposing eects are at work. In an open market there is always the risk that the second period market price is below 2 and in such a case the supplier does not sell anything at all. On the other hand, if the price exceeds 2 we have x 2L = 1. In a content protected market the local supplier (almost) always sells a positive amount, however as long as he charges p 2L he does not supply the nal good producer fully. Thus, in general the marginal eects of investment on expected prots in the second period may be smaller or larger under content protection. Furthermore, the marginal expected prot of investment also depends on the inuence of the investment on the survival probability. Thus, it depends on the parameter constellation under which market regime the optimal investment level without the risk of bankruptcy is larger. However looking at gure 1b) we realise that for 1 = 16 the relative weight of the two eects has ipped and the optimal investment level without credit restrictions is now larger without content protection. Whereas the investment level increases under both market regimes the increase is larger without content protection. Furthermore the policy change of the local supplier in the open market where he switches from > D to D occurs at a lower credit limit. We will not provide a complete analysis of the question under which circumstances content protection leads to higher investment in the absence of the risk of bankruptcy. Rather we will now concentrate on the eect of the initial eciency on the investment behaviour. We will carry out this analysis under two polar cases, namely that of unlimited credit and no credit. Let us rst discuss the optimal investment behaviour without content protection. The shape of the corresponding curve in gure 2a) can be easily understood taking into account our discussion above. Without any credit limit the optimal investment increases with 1 since the probability to be able to supply the nal good producer increases. On the other hand, if the credit limit is zero, we know from the previous paragraph that a very ecient rm should nevertheless invest a positive amount. The less ecient the rm is, the larger is the probability of bankruptcy and the smaller is the probability that the second period market price will be larger than 2. Together this implies that the marginal payo of the investment decreases 7 7 There is also a positive eect of an increase in 1 on the marginal payo of investment stemming from the fact that the marginal eect of on the survival probability increases with 1. However this eect is always dominated by the other two. 13

14 γ 1 γ 1 a) D = 1 b) D = 0 Figure 2: Optimal investment in open and protected (bold line) markets for dierent values of initial eciency 1 leading to a reduction of the optimal investment level. For initial eciencies larger than 19:5 ( 1 > 19:5) no investment is undertaken for D = 0. For the case of the protected market gure 2a) shows that investment is not strictly increasing with increased costs in period 1. This is due to a number of eects that inuence investment in the LC market. We have to realise that investment incentives are always larger under the assumption that the rm charges the market price in the second period than under the assumption that it charges p 2L. This is due to the fact that the survival probability is independent from the pricing decision in the second period and accordingly the marginal prot of investment only depends on the amount sold. Furthermore, a larger 1 corresponds to a larger probability to choose p 2L. Thus, although optimal investment increases with 1 given a second period pricing decision, there is an interval of 1 values where the optimal investment level decreases with 1. This interval corresponds to a range of values where the marginal eect of an increase of 1 on the probability to charge p 2L in period 2 is particularly large. For very ecient and very inecient rms optimal investment increases with 1 since the pricing decision in period 2 is almost deterministic and easy to predict. Considering gure 2b) we again see that the credit limit has basically no inuence on the optimal investment decisions on a protected market. Comparing the optimal level of investment with and without protection we see that without credit limit (D = 1) investment is slightly larger with 14

15 protection if the rm is extremely ecient. For a large range of 1 around the expected market price investment is substantially larger without protection. Only if the local suppliers' initial cost parameter is much larger than the expected market price, protection leads to a signicant increase in investment. For D = 0 the range of 1 where the eect of market protection on investment is negative is much smaller and entirely below the expected market price. Thus, whenever the initial costs of production lie above the expected market price and the local supplier can't borrow from the nancial markets (D = 0) content protection encourages investment, which in an open market would not occur. According to these results market protection should be used in such a case, which is exactly a scenario where content protection is used in actual markets (see section 1). Thus, in what follows we will focus on such cases and examine the eect of the industries' capital intensity on technological progress. This is in particular relevant since, as mentioned in the introduction, the UNIDO study from 1986 lists that most of the protected industries are highly capital intensive. To examine the impact of the capital intensity we vary between = 0:001 and = 0:09 and calculate the optimal investment for D = 0 and 1 = 21. The results are shown in gure ρ Figure 3: Optimal investment in open and protected (bold line) markets for 1 = 21 and D = 0. It can be seen that with increasing capital intensity (decreasing ) investment increases and that investment is always higher in the protected market. If the capital intensity reaches a critical value, which for the given parameter 15

16 constellation is = 0:061, investment only occurs in the protected market. At that point the expected payo of any investment falls short of the invested amount in case of the open market. Note that the gure implies that content protection is advantageous for any level of capital intensity. However the eect is much stronger if capital intensity is high. Hence our ndings are in accordance with empirical observations. Furthermore we show that even for sectors with low capital intensity local industries would prot from protection if only the eect on investment in process improvements is considered. 3.2 Optimal Investment under Dierent LC Schemes Having examined the eect of content protection per se in various settings, we will now turn to the question how the protection scheme should be designed to provide optimal investment incentives. A LC scheme in our model is dened by the two parameters and T. Hence, we will study the eects of variations of these two parameters. Again, we consider an input market in a highly capital intensive industry ( = 0:02) and a situation in which the local supplier can't borrow money from the nancial markets (D = 0). The LC requirement has a direct impact on the maximal price the local supplier may charge p jl and a direct as well as an indirect impact on the quantity. Holding the price constant, an increasing requirement causes the quantity supplied by the local supplier to rise. This leads to an increase of the costs the nal good producer has to bear when complying to the LC requirement. Since p jl is chosen such that the nal good producer is indierent between complying to, or violating the LC requirement, in order to maintain this status the local supplier has to lower p jl when is increased. This price adjustment leads to an indirect eect on the quantity supplied. However, since this indirect eect points in the same direction as the direct one, we can unambiguously state that an increase in results in a lower local content price and a larger quantity supplied by the local rm. The eect on the prot is however not clear. Concerning the incentives to invest we conclude that the increase in quantity yields an increase of the expected marginal payo of investment. Thus there should be a positive eect on investments. On the other hand, since prots stemming from the use of the LC price might increase with there can be a countereect on the investment incentives stemming from an increase of the probability that the LC price rather than the market price is charged in period 2. The tari rate T on the other hand has a direct eect on the maximal 16

17 price p jl but only an indirect eect on the quantity. For a xed price an increase in T does not change the quantity supplied. However it allows the local supplier to charge a higher price without the eect that the nal good producer chooses to violate the LC requirement. It follows from (8) that this leads to a lower quantity supplied by the local rm. Using a similar argument as in the previous paragraph we conclude that this should have a negative eect on the investment incentives. Since an increase in T enlarges the set of prices which will lead to a compliance with the LC requirement it is obvious that the eect on the local suppliers' prot has to be positive (if the LC price is chosen) or neutral (if the market price is chosen). Therefore the tari rate also has an impact on the local supplier's pricing decision. An increase in T increases the probability that the LC price is chosen. This again has two opposing eects on the investment incentives. An increase in the probability to choose the LC price reduces incentives, however the increase in the survival probability increases them. To examine the eects of dierent constellations of these two parameters set by government, we treat them separately by changing only one parameter at a time. Figure 4a) shows the results for dierent levels of the LC requirement if T = 0:2, gure 4b) depicts the results for T = 0: γ 1 γ 1 a) T = 0:2 b) T = 0:4 Figure 4: Optimal investment for dierent content requirements = 0:4 (bold line) and = 0:8 Let us rst consider the eect of an increase of. We can infer from gure 4b) that such an increase leads to an increase in investment when 17

18 the tari is rather high 8. For a low tari (gure 4a)) it depends on the initial eciency of the rm whether the eect is positive or negative. In such a case increasing has a positive eect only if the local supplier is neither extremely ecient nor extremely inecient. If the local supplier is very ecient, and with high probability chooses the market price in period 2 the eect that an increase in lowers this probability is dominant, reducing investment incentives. On the other hand, if 1 is very large, a large leads to a steep decrease in the rst period prots since the incentive for the nal good producer to violate the LC rule and buy nothing from the local supplier is large. Thus the survival probability of the local supplier decreases which implies decreasing investment incentives. This shows that it is important to choose and T in an appropriate ratio. In order to compare the eects of an increase in the tari rate we have to compare gures 4a) and 4b). From the solid lines it is easy to see that an increase in T leads to higher investment only for an inecient rm. For the ecient rm the eect of the investment on the probability to survive is negligible and thus increasing T decreases investment. For small values of obviously an increase in T always discourages investment. This has to be due to the fact that a decrease of increases the probability to survive and therefore the eect of an increase of T on the expected marginal eect of investment in cost reductions becomes more dominant. Finally let us analyse how the 'phase-out' of a LC scheme aects the rm's decision to invest. It is intuitive that protection of a market should stop as soon as the eciency of the local supplier is suciently high. Thus the question arises how the determination of content protection should be managed. If the eciency is already rather high one could argue that making the local supplier aware that the protection scheme will be weaker in the subsequent periods further encourages investment. Accordingly a slow phaseout of the content protection should be a sensible way to transform the protected market into an open one. Here we will examine in how far this intuition is supported by our model. To this end we compare the eects of constant protection with dierent phase-out policies. The results are shown in gure 5. In gure 5a) we let the initial LC scheme be T = 0:2 and = 0:8 while for gure 5b) we chose = 0:8 and T = 0:4, a setting in which, as stated above, it is always favourable for the local supplier to choose the LC 8 Note that for = 0:8; T = 0:4 we have T 1? section 2.1 that the local supplier always charges p jl. and can infer from our analysis in 18

19 price. In these settings we examined the eects driven by a decrease in both and T, compared with an unmatched decrease of a) T = 0:2 γ 1 b) T = 0:4 γ 1 Figure 5: Optimal investment if LC scheme remains unchanged (bold line), if LC requirement changes by 50%, and if both LC parameters change by 50% (dashed line) Looking at gure 5b) it can be seen that a phase-out scheme is counterproductive if initially a very high tari is used. The reason for this is that in such a case the survival probability { which is unaected by the phaseout { is large and the amount sold in the second period decreases sharply with decreasing. Furthermore, from a comparison of the two gures we conclude that as long as the eciency of the rm is very low a protection scheme with high, and over both periods constant values for both and T yields highest investment. On the other hand if 1 is rather low a scheme with a smaller initial value of T leads to better results. Here a phase-out policy is advantageous and we can see from gure 5a) that again it depends on the initial value of 1 whether only, or and T should be reduced in the second period. From these observations one can derive the following policy implications. Faced with a typical scenario where content protection is invoked (i.e. 1 is substantially larger than the expected world market price p jf ) a strong protection scheme should be used. This scheme should be upheld for a certain time to encourage the rm to make the necessary investments to close the gap between the own costs and the market price. However as soon as this gap becomes suciently small (but is still positive { see gure 6) the policy 19

20 γ 1 Figure 6: Optimal investment in protected markets if the LC scheme remains unchanged (bold line), if both parameters are decreased by 50% (dashed line), and if T is decreased by a further 50%. makers should reduce the tari and announce as a further weakening of the scheme, that will also be reduced in subsequent periods. Since the local rm is now already suciently ecient it expects to survive also under weaker conditions of protection given that it can further reduce costs. Due to this increased eciency the survival probability is rather insensitive to an increase in investment, while the expected sales in the subsequent period now become very sensitive with respect to costs, as the probability that the local rm supplies the nal good producer fully increases sharply. Therefore these phase-out policies lead to high investments and thus after a few periods the local rm's should be suciently ecient, such that the content protection can be abandoned altogether. 4 Conclusion In this paper we have contributed to the ongoing discussion on the usefulness of LC protection as a tool to facilitate the development of weak local industries. We have used a very simple two-period model to examine the eect of market protection on the investment incentives for local suppliers. From our simulation studies we have generated the following main ndings. Content protection is particularly useful, if rms can not borrow as much as they would like to borrow from the nancial markets. If there is no 20

21 credit limit the open market in general yields higher investment incentives. Furthermore we have conrmed the intuition that content protection is most advantageous if the local rms' costs are large compared to the expected world market price. A key impact on the usefulness of protection has the capital intensity of the protected industry. Using our model we could rigorously support the prevailing opinion that the eciency of market protection as a tool to encourage innovation is positively correlated with the capital intensity of the industry. We also compared dierent value added LC schemes with each other. We showed that increased protection in general leads to increased investment of the inecient rm. Furthermore our ndings imply that a good balance between tari and LC requirement is very important for the success of the scheme. Such a scheme should be upheld until the local supplier reaches a certain degree of eciency. At that point a slow phase-out policy should be used to abandon the protection. Following one of the standard assumptions of microeconomics we have assumed throughout the study, that all market participants are fully rational. Particularly the investment decisions of the local supplier were always assumed to be optimal in our model. However it has been argued by a number of scholars during the last decades, that the complete rationality assumption is too strong and does not provide a realistic picture of actual economic behaviour. Thus in order to validate our ndings about the usefulness of content protection we plan to complement this analysis by an examination of similar questions under the assumption that the local supplier acts boundedly rational. In particular we will assume that the local supplier uses an adaptive learning rule to determine his level of investment. Such an analysis will be the subject of future research. References [1] Grossman, G.M., "The theory of domestic content protection and content preference", Quarterly Journal of Economics, 96(4), 1981, [2] Hollander, A., "Content protection and transnational monopoly", Journal of International Economics, 23(3/4), 1987, [3] Krishna, K. and Itoh, M., "Content protection and oligopolistic interactions", Review of Economic Studies, 55(1), 1988,

22 [4] Lyon, T.P., "Regulatory hindsight review and innovation by electric utilities", Journal of Regulatory Economics, 7, 1995, [5] Mayo, J.W. and Flynn, J.E., "The eects of regulation on research and development: theory and evidence", Journal of Business, 61(3), 1988, [6] Munson, Ch. and Rosenblatt, M., "The impact of local content rules on global sourcing decisions", Production and Operations Management, 6(3), 1997, [7] Mussa, M., "The economics of content protection", NBER Working Paper no.1457, 1984 [8] Richardson, M., "The eects of a content requirement on a foreign duopsonist", Journal of International Economics, 31(1/2), 1991, [9] Sweeney, G., "Adoption of cost-saving innovations by a regulated rm", The American Economic Review, 71, 1981, [10] Teisberg, E.O., "Capital investment strategies under uncertain regulation", RAND Journal of Economics, 24(4), 1993, [11] UNIDO, Industrial policy in the developing countries: An analysis of local content regulations, UNIDO/IS.606, 1986 [12] Vousden, N., "Content protection and taris under monopoly and competition", Journal of International Economics, 23(3/4), 1987,