Illegal Immigration and Preferential Trade Liberalization. Subhayu Bandyopadhyay *

Illegal Immigration and Preferential Trade Liberalization Subhayu Bandyopadhyay * Department of Economics, West Virginia University Morgantown, WV-26506-6025, USA. April, 2001. Abstract This paper presents a version of the small-union Meade model to analyze the illegal immigration problem in the context of preferential trade liberalization. Two possible objectives for the host nation are explored: (i). holding the level of illegal immigration at an exogenously given social target; or, (ii). allowing for a variable illegal immigration level to maximize national welfare. The available policy instruments for the host nation are import tariffs/subsidies, border and internal enforcement. The second best tariff on the import from the partner country could either be positive or negative depending on the effect of the tariff on the wage rate and the pattern of substitutability in consumption. In scenario-(ii), greater enforcement that reduces labor inflow may contract the protected sector and confer additional benefits. If enforcement is too costly as a policy tool, tariff policy may be used to exploit monopsony power in the labor market and also to counter the distortionary effect of labor flows. ))))))))))))))))))))))))))))))))))))) * Please address all correspondence to: Subhayu Bandyopadhyay, Department of Economics, P.O. Box 6025, B & E Building, West Virginia University, Morgantown, WV-26506-6025 Tel: (304)-293-7879 Fax: (304)-293-5652 e-mail: bandysub@mail.wvu.edu

1. Introduction 1 The proliferation of regional trading agreements in recent years has revived substantial interest in the economics of preferential trading agreements. The literature has grown in this area to address issues such as optimality of the number of trading blocs, the conflicts or complementarities between multilateralism and bilateralism, and the political economy of preferential trading agreements. In several cases these trade blocs allow free trade in goods but restrict mobility of labor through immigration restrictions. For example, in spite of NAFTA, which has led to liberalization of trade between USA and Mexico, immigration restrictions remain and illegal immigration has emerged as a serious problem. 2 In general, trade agreements will affect output prices and in turn the wages in the member nations of the trading bloc. If the wage differential rises, illegal immigration will tend to rise and this will in turn impact enforcement policy and the possible gains from such trade agreements. While the issues of trade creation and trade diversion have been adequately addressed in the literature, I am not aware of any contribution in trade theory which addresses the effects of preferential trading agreements on illegal immigration, enforcement costs and national welfare. 3 This paper attempts to bridge these two separate strands in the literature and fill the gap. 1 I would like to thank Eric Bond and other participants of the mid-west international meetings at University of Kentucky for very helpful comments. 2 This issue is also pertinent to other nations. For example, there is extensive illegal immigration between Bangladesh and India which are SAARC (South Asian Association for Regional Cooperation) members. The desirability of forming a regional trading bloc of SAARC is being discussed. Preferential trading between India and Bangladesh will surely affect the illegal labor flows between these nations. 3 Some of the early contributions in the illegal immigration literature are Djajic (1987), Ethier (1986) and Bond and Chen (1987). The literature is growing rapidly, see for example, Djajic (1997), Gaytan- Fregoso and Lahiri (1998, 2000) and Hanson and Spilimbergo (1999a, 1999b). 1

From the literature on preferential trading agreements we know that trade blocs can be welfare enhancing or welfare reducing depending on the relative strengths of the trade creation and diversion effects. Panagariya (1997) among others uses a small-union Meade model to address these effects and derives the second best optimal tariff on intra bloc trade in the presence of an existing import tariff on a third nation. If this second best tariff is positive (due to trade diversion), a preferential trading agreement that leads to its elimination has an ambiguous welfare effect. In addition, if the small country assumption is discarded, terms of trade effects of preferential liberalization can play an important role. A trade bloc benefits members by improving their terms of trade vis-a-vis non-members, but some bloc members may suffer from adverse terms of trade effects of intra-bloc liberalization. Riezman (1979) shows that a sufficient condition for the first of these effects to dominate (ensuring that all members benefit) is that intra-union trade is small. The literature on illegal immigration has looked at enforcement policy on the basis of two possible objectives. In Ethier (1986) the focus is on choosing an efficient mix of border and internal enforcement to target an exogenously given (from social considerations) illegal immigration level. On the other hand, Bond and Chen (1987) focus on the wage tax effect of internal enforcement and how that may raise national income (and welfare). 4 This paper considers the two issues of preferential trade liberalization and illegal immigration simultaneously in a general equilibrium context. Using a small-union Meade model we address optimal policy combinations of internal enforcement, border enforcement and the second best import tariff (on the member nation) that will achieve welfare maximization under: 4 This line of investigation is in a similar vein as the literature on optimal restrictions on labor/capital flows pioneered by Ramaswami (1968) among others (also see Jones, Coelho and Easton, 1986). 2

(a). a target illegal immigration level; (b). a variable illegal immigration level. The analysis of second best tariff policy informs us about the possible gains from preferential tariff liberalization. For example, if labor mobility tends to reduce the second best tariff, the potential for a welfare augmenting trade bloc (that pursues intra-bloc free trade) is increased. The discussion of simultaneous use of enforcement policy and tariff policy and the interactions between trade and factor mobility that they capture complement the existing literature in both these areas. 2. The Model We use a small-union Meade model (similar to Panagariya, 1997) where three nations A, B and C trade with each other. A and B are small in relation to C, so that the border prices are determined in nation-c. A and B can affect their domestic prices by using import tariffs. There are three goods 1,2 and 3. Goods 1 and 2 are produced and consumed by A and B while they only consume good-3 and do not produce it. A exports good-1 and imports goods 2 and 3. B exports good-2 and imports goods 1 and 3. C exports good-3 and imports goods 1 and 2. We discuss preferential trade liberalization by A in terms of its reduction of import tariff on B (i.e., on good-2). Since border prices are given exogenously in this model we normalize them to equal unity. The tariff rate of nation-a on good-2 is t A 2 and that on good-3 is t A 3. Similarly, nation-b s tariffs are t B i (i=1,3). The wage rate in nation-a (w A ) is assumed to exceed the wage in nation-b (w B ). Illegal immigration occurs in response to this wage difference between B (the source nation) and A (the host nation). 5 Let e i and e b be the 5 If we allow illegal immigration between C and A the central results are not qualitatively altered. The elasticity of supply of labor will be different and will alter the effects of tariff changes or enforcement on illegal immigration (see equations 4a through 4e below). However, since the effects of all the policy tools are affected symmetrically by this change in elasticity, the relative effects are not altered. The 3

internal and border enforcement levels used by A to detect illegal immigrants (see Ethier, 1986 and Bond and Chen, 1987). The probabilities of internal and border detection are denoted by p i and p b, respectively. The probability functions are assumed to be: p i = p i (e i ), p i N>0, p i O<0; p b = p b (e b ), p b N>0, p b O<0. (1) Let I be the level of illegal immigration, and v i and u i be the endowment vector and welfare level of nation-i (i=a,b). Let w i be the wage rate in nation-i and w I be the illegal wage rate in the host nation. The expenditure-revenue relationship for nation-a is: 6 E A (1, 1+t A 2, 1+t A 3, u A ) = R A (1, 1+t A 2, v A + I) + t A 2(E A 2 - R A 2) + t A 3E A 3 - w I I - e i -e b. (2) Like Bond and Chen (1987) we assume that firms are fined z per unit of illegal labor detected. Thus, the expected fine per illegal worker hired is zp i. The cost of hiring a unit of illegal labor is the sum of this expected fine and the illegal wage rate w I. Competitive firms will equate the cost of hiring a unit of illegal labor to the cost of hiring legal labor. This yields: w A = w I + zp i. (3) The probability of being caught at the border is p b and f is the cost for the immigrant who is detected and returned to her source country (see Ethier, 1986). If successful at the border then the immigrant finds employment at a wage w I in nation-a. Assuming risk neutrality a potential immigrant equalizes the expected wage from illegal immigration to the certainty wage in B: central results of this paper depend on the relative effects of the policy tools in controlling illegal immigration and are therefore unchanged. 6 Since good-3 is not produced in A or B, its price does not feature in the revenue function of these nations. Also, in the tradition of the existing literature we assume that income of illegal immigrants is not considered to be a part of A s national income. 4

w B = w I (1-p b ) + (w B - f)p b Y w B = w I - fp b /(1-p b ). (4a) Using (1), (3) and (4a) and denoting p b /(1-p b ) as the function è(e b ) we have: w B = w A - zp i (e i ) - fè(e b ). (4b) Further, noting that the partial of the revenue function with respect to factor endowment yields the factor reward vector, and using (4b) we have: R B v(1+t B 1, 1, v B -I) - R A v(1,1+t A 2, v A + I) + zp i (e i ) + fè(e b ) = 0. (4c) Using (4c) we find that: I = I(e i, e b, t A 2, t B 1). (4d) We assume that the revenue function is strictly concave in v for both nations such that R i vv < 0 (i=a,b). 7 Let D = R A vv + R B vv <0. Denoting the partial derivative of the I(.) function with respect to the j-th. argument by I j (.) we have from (4c) and (4d): I 1 = zp i N/D<0; I 2 = fèn/d<0; I 3 = -R A v2/d, I 4 = R B v1/d. (4e) Thus the effect of enforcement is to reduce illegal immigration (I 1 and I 2 are negative) but the effect of tariff changes are ambiguous in general and depend on their effect on the factor rewards in each nation. For example, if a tariff increase in nation-a on good-2 leads to a higher wage rate (i.e., R A v2>0) it must raise illegal immigration (I 3 >0). Of course R A v2 is the same as R A 2v and thus represents (as well) the effect of illegal labor flows on output of sector-2. Similarly, a change in nation-b s tariff rate will affect 7 See Miyagiwa and Young (1986) for a similar assumption regarding the revenue function in a related context. Of course, the Heckscher-Ohlin framework is not consistent with such a framework. A specific factors model, where endowment changes have an effect on factor rewards even when output prices are constant is more appropriate for analyzing the effects of illegal labor mobility (see Bandyopadhyay and Bandyopadhyay, 1998, for example). 5

w B and change the incentives for migration and raise or lower illegal immigration appropriately. Using (2) and (3): E A (1, 1+t A 2, 1+t A 3, u A ) = R A (1, 1+t A 2, v A + I) + t A 2(E A 2 - R A 2) + t A 3E A 3 - w A I - e i -e b + p i (e i )Iz. (2N) Totally differentiating (2N) we find that: E A udu A = t A 2dM A 2 + t A 3dE A 3 - Idw A - de i - de b + Izdp i + p i zdi; M A 2 = E A 2 - R A 2. (5) Because of existing tariff distortions in goods 2 and 3, any rise in imports (M A 2 and E A 3 are imports of goods 2 and 3, respectively) is a movement in the direction of free trade and is welfare enhancing. 8 Also, note that these gains are weighted by the level of the existing tariff distortion t A i (i=2,3). The third term in the right hand side represents the loss from a higher host nation cost of hiring an illegal worker, the fourth and fifth terms the enforcement costs and the last two the change in fine collections from the employers (i.e., d{p i (e i )Iz}). 9 The literature has analyzed two host country objectives for controlling illegal immigration: (i). holding it at a socially desirable target level; (ii) choosing it at a level that maximizes national income. We consider the first objective in the following sub-section. 2.1 Target Immigration Level: I = & I, di = 0. Here & I is some exogenously given social target. From (4d) we can infer that there are three policy instruments (e i, e b, t A 2) at the disposal of A to target this immigration level. Suppose that these 8 See Berglas (1979), Panagariya (1997) among others for analogous findings. 9 The fines collected from the employers are passed on to the illegal workers in the form of a lower w I and thus acts as an effective wage tax on illegal workers (see Bond and Chen, 1987). Given w A, a rise in this fine collection does add to the welfare of nation-a. 6

instruments are chosen appropriately to maintain I at the target level. Then (5) reduces to: E A udu A = t A 2dM A 2 + t A 3dE A 3 - Idw A - de i - de b + Izdp i. (5N) Given t A 3, assuming normality of all goods, (5N) yields: Sdu A = {t A 2(E A 22 - R A 22) + t A 3E A 32}dt A 2 - I(dw A - zdp i ) -de i -de b ; S = E A u - t A 2E A 2u - t A 3E A 3u = E A u1 + E A u2 + E A u3 > 0. (6) Noting that w A equals R A v(1,1+t A 2, v A + I), (6) reduces to: Sdu A = {t A 2(E A 22 - R A 22) + t A 3E A 32 - IR A v2}dt A 2 -de b - de i (1-zIp i N). (7) Utilizing (4d) and (4e) and the fact that di=0, we have (for a given t B 1): de b = (R A v2/fèn)dt A 2 - (zp i N/fèN)de i. (8) Using (8) in (7): Sdu A = [t A 2(E A 22 - R A 22) + t A 3E A 32 - R A v2{i + (1/fèN)}]dt A 2 + {(zp i N/fèN) - (1-zIp i N)}de i. (9a) Assuming that internal enforcement is optimally chosen: zp i N/fèN = 1-zIp i N Y I + (1/fèN) = 1/zp i N. (9b) Using (8) it is clear that the first term in (9b) yields the marginal rate of substitution (MRS) between internal and border enforcement for a given I. On the other hand, consider the net cost of enforcement as e i + e b - p i zi &. The slope of an iso-cost in (e i,e b ) space is: 1-zIp i N. Thus, (9b) ensures that MRS equals the slope of an iso-cost, and therefore corresponds to the point of minimum cost for attaining the target iso-immigration curve. Using (9b) in (9a): Sdu A = [t A 2(E A 22 - R A 22) + t A 3E A 32 - (R A v2/zp i N)]dt A 2. (10a) (10a) yields the second-best optimal tariff on good-2 (assuming that t A 3 is given): t A 2 = {t A 3E A 32 - (R A v2/zp i N)}/(R A 22 - E A 22). (10b) 7

Convexity of the revenue function and concavity of the expenditure function implies that the sign of the second best tariff on good-2 depends on the numerator of the right hand side of (10b). Assuming that goods 2 and 3 are net substitutes, the tariff distortion on good-3 tends to raise the second best tariff. On the other hand, a positive R A v2 tends to reduce the second best tariff on good-2. The balance of these effects determine the magnitude and sign of the second best tariff. Proposition-1 The second best tariff on good-2 must be positive (negative) if goods 2 and 3 are net substitutes (complements) in consumption for nation-a, and if a rise in the tariff on good-2 reduces (raises) the wage rate in nation-a. Otherwise, the sign of the tariff is ambiguous and depends on the relative magnitude of the cross substitution effect and the wage effect. Proof and Comment Relation-(10b) provides the proof. The term t A 3E A 32 reflects the distortionary effect on good-3 of a fall in the tariff on good-2. A fall in t A 2 will tend to reduce the demand for good-3 under substitutability. Given that there is already under-consumption of the good due to the existence of a positive t A 3, this increases the distortion in good-3. This effect justifies a higher tariff on good-2. 10 On the other hand, if R A v2 is negative, a fall in t A 2 will raise the wage rate in nation-a and raise the incentive for illegal immigration. Thus, more enforcement will be necessary to maintain the same target level & I (recall equation-4d and 4e). 11 This rise in enforcement cost reduces welfare. Therefore, this second effect as 10 See for example the small-union Meade model in Panagariya (1997). 11 Note from (4e) that Me i /Mt A 2 = -I 3 /I 1 = R A v2/zp i N. If R A v2<0, then the iso-immigration curve in (e i, t A 2) space is negatively sloped and (R A v2/zp i N) measures the marginal rate of substitution between internal enforcement and the tariff on good-2. 8

well tends to raise the second best tariff on good-2 (if R A v2<0). Thus, if E A 32 is positive and R A v2 is negative, the second best tariff on good-2 is necessarily positive. The other cases may be explained similarly. The literature has recognized that if the second-best tariff is positive then a preferential trade liberalization that leads to a zero tariff on good-2 may not be welfare enhancing. Essentially, trade diversion costs may outweigh the trade creation gains. We complement this result by noting that if R A v2<0, the second best tariff tends to rise in the presence of illegal immigration. Thus a preferential movement to free trade in good-2 is even less likely to raise welfare. The enforcement costs compound the trade diversion costs of preferential liberalization leading to this rather pessimistic scenario. However, if R A v2>0, the enforcement costs are lowered because of the preferential liberalization and this will tend to raise the desirability of such an agreement. Finally, note from (10b) that even if there is free trade in good-3, free trade in good-2 is not optimal in general. At free trade (t A 2 = t A 3=0) the tariff distortions are weighted by the respective tariff rates and tend to zero. Using (4e), at free trade (10a) reduces to: Sdu A = (- R A v2/zp i N)dt A 2 = (-Me i /Mt A 2)dt A 2. (10c) The marginal rate of substitution between enforcement costs and the tariff on good-2 does not vanish at free trade. If R A v2<0, a marginal rise in t A 2 will reduce w A and allow a reduction the enforcement costs and justify a tariff. Based on (10b) we have (at t A 3=0): t A 2 = - (R A v2/zp i N)/(R A 22 - E A 22). (10d) An internal optimum is reached where the tariff distortion in good-2 is exactly balanced by the marginal 9

effect of the tariff on enforcement costs. 12 2.2 Optimal Policy under a Variable Immigration Level. This section focuses on the choice of available policy instruments to maximize welfare in the host nation allowing for a variable immigration level. The focus is closest to Bond and Chen (1987) with the major differences being that this is being done in the context of a preferential trading model and that we allow for border enforcement along with internal enforcement. The analysis also complements the literature on optimal restrictions on factor mobility as in Ramaswami (1968) and Jones et al. (1986) among others. A major difference with these latter papers is that we are discussing illegal labor flows and therefore its level cannot be directly chosen - it can only be indirectly influenced through the choice of the available policy tools of internal and border enforcement and tariffs. From (5) above we can derive: Sdu A = (X - IR A v2 + YI 3 )dt A 2 + {YI 1 - (1-Izp i N)}de i + (YI 2-1)de b ; where, X = t A 2(E A 22 - R A 22) + t A 3E A 32; Y = p i z - IR A vv - t A 2R A 2v. (11) The marginal returns on border enforcement are: 12 Since the effect on enforcement costs is through the effect of the price of good-2 on the wage rate, a first best policy is a production subsidy (or tax) on good-2 rather than an import tariff (or subsidy). If R A v2<0, a production subsidy reduces w A and reduces the incentive to immigrate and thus enforcement may be lowered for a target I. The saving in enforcement expenditure raises A s welfare just as in the tariff case. The efficiency cost of this policy (which only distorts production) is however lower than that for an import tariff which distorts consumption as well. We focus on tariffs because of two reasons: (i) the second best tariff is important not because we believe that it is being chosen, but rather to use it as a benchmark for identifying potential welfare improvements from preferential trade liberalization; (ii) while a production tax/subsidy may be economically efficient, they are in general politically difficult to implement (a production subsidy program has to apply to the entire volume of domestic production of the good; therefore it will require a large outlay which has to be financed by taxation which will be unpopular). 10

S(Mu A /Me b ) = YI 2-1. (11a) Recall from (4e) that I 2 is negative. If t A 2 is zero, Y must be positive and Mu A /Me b <0 and thus border enforcement is zero. Border enforcement hurts on all three counts (under free trade in good-2): first, the cost of the enforcement itself; second, the reduction in I that it induces leads to a loss in fine collections to the tune of p i z per unit (first term in Y), and finally, the reduction in I leads to a leftward shift in the domestic supply of unskilled labor raising w A and hence the cost of hiring illegal labor to the firms (the second term IR A vv in Y). However, there is a potential benefit from border enforcement in the presence of a tariff on good-2. If R A 2v is positive, an inflow of illegal labor raises the production of good-2 and accentuates the production distortion due to the tariff (i.e., the term t A 2R A 2v in Y captures this effect). The negative effect of border enforcement on I helps in reducing the production distortion (assuming R A 2v > 0) and confers a welfare gain. If this gain is large enough, a positive level of border enforcement may be justified. On the other hand, the marginal benefit from internal enforcement is captured by: S(Mu A /Me i ) = YI 1 + Izp i N - 1. (11b) The first term on the right hand side of (11b) captures effects that are similar to those discussed above for border enforcement and the last term of (11b) captures the direct cost of internal enforcement. The second term on the right hand side is qualitatively different from the case of border enforcement and reflects the wage tax effect of internal enforcement identified by Bond and Chen (1987). Recall from (3) that zp i is the wedge between the legal and illegal wage and also note that this is collected by the government in A as revenue. Internal enforcement raises this wedge by zp i N (per unit of I) at the margin. This shows up as additional collections by the government of A to the tune of Izp i N. In the 11

absence of trade in goods (as in Bond-Chen) or in the absence of a tariff on good-2 in our model, Y is necessarily positive. While border enforcement is necessarily zero in this case, internal enforcement can still be positive because of the wage taxation effect. Next, let us consider the three possibilities of: (a). interior solutions for both internal and border enforcement; (b). interior solution for internal enforcement only; (c). corner solutions for both border and internal enforcement. In the first of these cases, the first order conditions for the choices of internal and border enforcement yield: Y = 1/I 2 ; and, Y = (1-Izp i N)/I 1 Y I 1 /I 2 =1-zIp i NY zp i N/fèN = 1-zIp i N. (12) Using (12) in (11) we obtain: S(Mu A /Mt A 2) = t A 2(E A 22 - R A 22) + t A 3E A 32 - (R A 2v/zp i N). (13) Comparing (13) and (10a) we find that regardless of whether immigration is variable or at a target level the marginal effect of tariff policy is qualitatively similar. Consequently, the second best tariff formula is the same in the two cases. However, the level of the second best tariff is different in general because the level of I, e i and e b are all endogenous here and affects the level of the tariff. If border enforcement is zero, then using the optimality condition for internal enforcement we can derive (13) again. Consequently, the second best optimum tariff is unaffected in qualitative terms. Finally, if both types of enforcement are zero, then we obtain: S(Mu A /Mt A 2) = t A 2(E A 22 - R A 22) + t A 3E A 32 - t A 2R A 2v(MI/Mt A 2) + IR B vv(mi/mt A 2). (14) Using (4e) and (14): S(Mu A /Mt A 2) = t A 2(E A 22 - R A 22) + t A 3E A 32 + {t A 2(R A 2v) 2 /(R A vv + R B vv)}+ IR B vv(mi/mt A 2). (14N) The third term in the right hand side of (14N) is the same as -t A 2R A 2v(MI/Mt A 2) in (14) and is unambiguously negative. Factor mobility induced by a tariff increase accentuates the existing production distortion 12

regardless of the sign of R A v2. Recall from (4e) that: MI/Mt A 2 = -R A v2/(r A vv + R B vv). Suppose that R A v2>0 and thus MI/Mt A 2 is positive. As the tariff on good-2 is raised, w A and I will rise, in turn raising production of good-2 (since R A v2=r A 2v >0). The expansion of the protected sector is welfare reducing. Similarly, if R A v2<0, a rise in the tariff will reduce w A and I, and thereby raise the production of good-2 (since R A 2v<0). Thus, in either case, the factor mobility implications of the tariff on good-2 are harmful in terms of production distortions. This tends to reduce the magnitude of the second best tariff on good-2. If R A v2 is negative, (4e) implies a tariff on good-2 will reduce I. The increase in labor supply in nation-b reduces w B (since R B vv<0). As enforcement is zero, w B equals w A and the lower wage in A reflects lower payment to illegal labor. The illegal wage payments fall in aggregate to the extent of IR B vv. The effect of the tariff here is to exploit the monopsony power in the labor market [captured in the last term of (14) and (14N)]. From (14N), we get the second best tariff on good-2: t A 2 = [t A 3E A 32 + IR B vv(mi/mt A 2)]/D * = [t A 3E A 32 - {R A v2ir B vv/(r A vv + R B vv)}]/d *, where, D * = R A 22 - E A 22 - {(R A 2v) 2 /(R A vv + R B vv)} > 0. (15) The tariff tends to rise above the traditional level (assuming net substitutability in consumption) to exploit the monopsony power in the labor market if R A v2 is negative. On the other hand, the effect of factor mobility is to raise the production distortion, the tariff falls to compensate for that (captured by a larger D * ). However, as long as R A v2 is negative, the tariff must be positive (under substitutability in consumption). If, on the other hand R A v2 is positive, the level of the second best tariff will be reduced compared to the case where there is no illegal immigration. Proposition-2 13

A tariff distortion on good-2 may justify a positive level of border enforcement. Even if border enforcement is zero, internal enforcement may still be positive. For positive enforcement levels, the second best tariff on good-2 is similar to the case of targeted immigration. If both internal and border enforcement are zero, a second best tariff on good-2 must reflect (in addition to standard efficiency losses): (i). the effect of illegal immigration on production distortion in the protected sector, and, (ii). the effect of illegal immigration on the illegal wage (i.e., the monopsony power). Proof and Comment The proof and explanation is contained in the above discussion. We summarize the findings here. If the purpose of controlling illegal immigration is based solely on economic efficiency grounds, border enforcement cannot be justified based on the existing work. This is because it uses up resources, raises illegal wage in the host nation by reducing the supply of labor and reduces the fines collected (zp i I) because of a reduction in I itself. However, under import tariffs, a reduction in illegal immigration may shrink the protected sector and justify its presence. Even under free trade, internal enforcement may be positive as long as the wage tax effect that it utilizes [recall Izp i N in (11b)] is strong. If there is positive enforcement (border or internal), the second best tariff level is qualitatively similar to the first case where I is held at a target level and performs a similar function. It is possible that on efficiency grounds neither internal or border enforcement may be justified. In this scenario, the tariff on good-2 has to substitute for enforcement policy in extracting the gains from both monopsony power and a possible reduction in production distortion in the protected sector. On both of these grounds if R A 2v is positive, the second best tariff on good-2 is reduced. 3. Conclusion 14

This paper provides a framework for analyzing the problem of illegal immigration in the context of (bilateral or multilateral) trade liberalization. For ease of exposition we have focused on a limited number of issues. However, the framework is quite general and may be easily adapted to addressing other issues involving trade and factor mobility. We find that the second best tariff on a member nation is dependent on both the tariff on the non-member and the effect of the second best tariff on factor reward. If the second best tariff is reduced due to the immigration considerations then there is a greater possibility that a preferential trading agreement like an FTA or CU is beneficial. Optimal enforcement policy is not independent of trade considerations. Some new results emerge regarding optimality of enforcement when some sectors are protected. In particular, greater enforcement may reduce the availability of labor and help in shrinking the protected sector. In addition to controlling illegal immigration, this trade related efficiency gain may justify greater enforcement. References: Bandyopadhyay, S. and S.C. Bandyopadhyay, 1998, Illegal immigration: A supply side analysis, Journal of Development Economics, 57, 343-360. Berglas, E., 1979, Preferential trading theory: The n commodity case, Journal of Political Economy, 87(2), 315-331. Bond, E.W. and T.J. Chen, 1987, The welfare effects of illegal immigration, Journal of International Economics, 23, 315-328. Djajic, S., 1987, Illegal aliens, unemployment and immigration policy, Journal of Development Economics, 25, 235-249. Djajic, S., 1997, Illegal immigration and resource allocation, International Economic Review, 38(1), 15

97-117.. Ethier, W.J., 1986, Illegal immigration: The host country problem, American Economic Review, 76, 56-71. Gaytan-Fregoso, H. and S. Lahiri, 1998, Controlling illegal immigration: Border control vs. domestic enforcement, working paper, Dept. of Economics, University of Essex. Gaytan-Fregoso, H. and S. Lahiri, 2000, Foreign aid and illegal immigration, Journal of Development Economics, 63, 515-527. Hanson, G.H. and A. Spilimbergo, 1999a, Illegal immigration, border enforcement, and relative wages: Evidence from apprehensions at the U.S.-Mexico border, American Economic Review, 89(5), 1337-1357. Hanson, G.H. and A. Spilimbergo, 1999b, Political economy, sectoral shocks, and border enforcement, NBER working paper#w7315. Jones, R.W., I. Coelho and S.T. Easton, 1986, The theory of international factor flows: The basic model, Journal of International Economics, 20, 313-327. Miyagiwa, K.F. and L. Young, 1986, International capital mobility and commercial policy in an economic region, Journal of International Economics, 20, 329-341. Panagariya, A., 1997, The Meade model of preferential trading: History, analytics and policy implications, working paper, University of Maryland. Ramaswami, V.K., 1968, International factor movement and the national advantage, Economica, 35, 309-310. Riezman, R, 1979, A 3x3 model of Customs Unions, Journal of International Economics, 9, 341-16

354. 17