International Migration, Purchasing Power Parity (PPP) and the Money Metric of Welfare Gains

Transcription

1 International Migration, urhasing ower arity ( and the Money Metri of Welfare Gains ans immer and Dominique van der Mensbrugghe World Bank 1818 Street W Washington D htimmer@worldbank.org dvandermensbrugg@worldbank.org May 1, 2006 aper prepared for the 9 th nnual onferene on Global Eonomi nalysis osted by the United ations Eonomi ommission for fria (UE ddis baba, June 15-17, 2001

2 International Migration, urhasing ower arity ( and the Money Metri of Welfare Gains ans immer and Dominique van der Mensbrugghe May 1, 2006 bstrat: his paper argues that the rise in real global inome as a result of migration from low-inome ountries to high-inome ountries overestimates the money metri of global welfare gains by at least a fator of two. his is ontrary to the analysis of trade poliies where the rise in real inome, alulated aording to national aounting onvention, does provide an aurate estimate of the money metri of welfare gains. he ompliation in migration senarios arises beause the prie hanges that a migrant experienes, as pries of non-tradable goods differ signifiantly between the host and home ountries, are not represented in the deflators of the host ountry, but should be taken into aount in the money metri of the welfare hange of the migrant. djustment of real inome of migrants using purhasing-powerparity ( adjustment fators provides a good approximation of the money metri of welfare gains. he orret appliation of the money metri of welfare produes in the ase of migration a ounterintuitive example of money transfers that inrease the money metri of global welfare. Remittanes by migrants bak to their home ountry should not be -adjusted, as those who remain in the host ountry, do not experiene a signifiant hange in pries. Beause inome of new migrants that is not remitted home should be -adjusted, remittanes by new migrants inrease global welfare. ans immer, Manager, and Dominique van der Mensbrugghe, ead Eonomist, Development rospets Group, he World Bank. he authors gratefully aknowledge disussions with and suggestions from François Bourguignon, Uri Dadush and lan Winters. he views and findings in this paper are those of the authors and not neessarily those of the World Bank, its Exeutive Board, or member ountry governments. - ii -

3 able of ontents 1. IRODUIO MOEY MERI D EUIVE OR OMESIG VRIIO IIO USIG E WORD BK S IKGE MODE OUSIOS iii -

4 1. Introdution oliies that redue obstales to migration will hange global welfare. Many people will enjoy gains in welfare, while some will suffer losses; in the same way the elimination of trade barriers generates winners and losers. While eonomi models provide insight in gains or losses for a typial onsumer or worker, aggregation of the welfare gains and losses is not straightforward, beause welfare is fundamentally inomparable aross persons. ow an we determine if the gain for one person an more than ompensate the loss for another person? Standard pratie in applied general equilibrium modeling is to represent a hange in welfare by a hange in inome. For eah deision making unit an equivalent variation (EV in inome is omputed that would have generated at the original pries the same inrease in welfare as the atual inrease in welfare in a simulation. So, all the hanges in inome and pries in the simulation are summarized in one money metri. hese EVs are then aggregated over persons to alulate the hange in global welfare. he logi behind this global aggregate is that the hanges in welfare observed in a simulation ould have been ahieved at onstant pries if the global inrease in inome would have inreased by the amount of the aggregate EVs, aompanied by the appropriate inome redistribution. So, if global welfare inreases in terms of EVs, it does not mean that the welfare gains of the winners are onsidered more important than the welfare losses of the losers, but that the shok is similar to a situation in whih global inome would have inreased and redistribution of inome would have taken plae. good approximation of the EVs is the inrease in real inome. he approximation is so good that often only real inome is alulated in modeling exerises to alulate global welfare gains. owever, the real inome gains as alulated in ational ounts, whih basially apply to ountries and not to individuals, are no longer relevant one persons move from one ountry to another. herefore, we have to rethink the measurement of EV for migrants. he main purpose of this paper is to desribe a new way of alulating the equivalent variation of inome for migrants. 2. Money metri and equivalent or ompensating variation oliies that redue obstales to migration will hange global welfare. Many people will enjoy gains in welfare, while some will suffer losses; in the same way the elimination of trade barriers generates winners and losers. While eonomi models provide insight in gains or losses for a typial onsumer or worker, aggregation of the welfare gains and losses is not straightforward, beause welfare is fundamentally - 1 -

5 inomparable aross persons. ow an we determine if the gain for one person an more than ompensate the loss for another person? Standard pratie in applied general equilibrium modeling is to represent a hange in welfare by a hange in inome. For eah deision making unit an equivalent variation (EV in inome is omputed that would have generated at the original pries the same inrease in welfare as the atual inrease in welfare in a simulation. So, all the hanges in inome and pries in the simulation are summarized in one money metri. hese EVs are then aggregated over persons to alulate the hange in global welfare. he logi behind this global aggregate is that the hanges in welfare observed in a simulation ould have been ahieved at onstant pries if the global inrease in inome would have inreased by the amount of the aggregate EVs, aompanied by the appropriate inome redistribution. So, if global welfare inreases in terms of EVs, it does not mean that the welfare gains of the winners are onsidered more important than the welfare losses of the losers, but that the shok is similar to a situation in whih global inome would have inreased and redistribution of inome would have taken plae. good approximation of the EVs is the inrease in real inome. he approximation is so good that often only real inome is alulated in modeling exerises to alulate global welfare gains. owever, the real inome gains as alulated in ational ounts, whih basially apply to ountries and not to individuals, are no longer relevant one persons move from one ountry to another. herefore, we have to rethink the measurement of EV for migrants. he main purpose of this paper is to desribe a new way of alulating the equivalent variation of inome for migrants. Before we introdue this new alulation of EV for migrants, we first define the onepts in a more formal way and we introdue a very simple two-ountry model. he purpose of the model is not to be omprehensive or realisti, but to larify some basi onepts. fter the new way to alulate EV for migrants is introdued in this small model, the seond part of the paper will apply the new onept to the World Bank s global IKGE model, whih provides a more omprehensive and more realisti illustration of the onsequene of the new approah. ssume two ountries, a high-inome ountry and a low-inome ountry. In eah ountry onsumers an purhase two goods, a non-tradable good and a tradable good. hey make their onsumption deisions on the basis of a obb-douglas utility funtion. For simpliity we assume that all people have the same utility funtion, irrespetive of whether they live in ountry or ountry

6 U θ, Where U Utility of representative onsumer in ountry. s amount onsumed per person of produt s in ountry ll inome is onsumed and demand for the two produts is thus with Y (1 Y n + t 1 Where Y is per apita inome in ountry. Substitution of these demand equations in the utility funtion gives the indiret utility funtion U with Y / 1 n θ 1 1 he money metri M of utility level U at pries and is M ( U,, U If a shok in a simulation has generated a new welfare level, the equivalent variation in inome an be alulated as EV M ( U Old ew EW Y Y, ew Old OD, OD 1 M M ( U OD ( U, OD OD, OD, OD, OD If we hoose units of measurement suh that in the initial situation both pries are equal to 1 and if Old inome is 1, then the level of indiret utility isu 1. ssume the prie of non-tradable goods jumps in a simulation to 10, but inome does not hange. Indiret utility is then redued to new U 10. he - 3 -

7 money metri of that new utility level at old pries is new M 10. nd the equivalent variation (EV in inome of the prie inrease is EV M new M old 10 1 his equivalent variation is exatly the inrease in real inome if we use as deflator 1 θ 1 o further illustrate the use of equivalent variation, we add prodution to the model and look at the impat of aid flows. Eah ountry produes both goods with only labor as input. abor produtivity in the nontradable setor does not differ between the two ountries, but the labor in ountry is ten times as produtive as labor in ountry in the prodution of tradable goods ,, Where s labor used in setor s and in ountry total labor fore in ountry. s amount produed of produt s in ountry In the initial situation without aid or other transfers there is no trade between the two ountries. Eah ountry onsumes the amount of tradable and non-tradable goods it produes and for simpliity we assume that the whole population works. he inlusion of dependant hildren or elderly would not hange the basi analysis. / s s If we take the tradable goods as numéraire, pries are Where prie of tradable goods prie of non-tradable goods in ountry

8 Under these assumptions labor will be alloated similarly in the two ountries over the two setors: ( 1 er person, both ountries will produe and onsume the same amount of non-tradable goods. owever, ountry will produe and onsume per apita 10 times as many tradable goods. (1 10(1 (1 lthough only produtivity in the tradable goods setor is larger in ountry, wages and per apita inome in ountry are 10 times wages and per apita inome in ountry. Y Y ( ( + + / / 1 What is the impat of aid provided by ountry to ountry? et s assume people in ountry give a share of their inome to people in ountry. he demand equations beome now EW EW (1 Y / (1 (1 Y / s tehnology is fixed, pries will not hange, but people in ountry will onsume now less of both goods. hey will produe less of the non-tradable good, but they will produe more of the tradable good in order to export tradable goods to ountry. EW EW (1 ( (1 + Where the following represents the amount of traded goods exported: E

9 he money metri of the welfare loss of a person in ountry is ID EV 10 otal welfare loss in ountry aording to this measure is EV 10 nd one again this is exatly the deline in real inome in ountry. In ountry, onsumption of both goods will inrease EW EW 1 + ( ( If ountry ontinues to produe both goods (see below for the opposite ase, then also there pries will not hange, but resoures will be shifted away from the prodution of tradable goods towards the prodution of non-tradable goods. EW EW ( + 10 EW (1 ( + (1 10 he equivalent variation desribing the welfare gains is equal to the inrease in real inome for ountry. EV 10 In this example aid does not hange global inome, or the money metri of global welfare. If aid is large enough to push ountry ompletely out of the prodution of tradable goods, then pries are no longer fixed. Inreased demand for non-tradable goods will push pries of those goods up. dditional aid beyond that point only has a limited impat on welfare and real inome in the reeiving ountry. he EV-gain in the reeiving ountry of an additional aid dollar is less than the EV-loss in the donor ountry and will thus redue global real inome. In formulas: It must be that is zero, whih implies that: (1 (1 (

10 - 7 - It is also the ase that export of tradables equals aggregate onsumption, therefore the following must hold: + a E with per apita inome in the region is equal to the wage plus the value of aid per apita. From this identity we an dedue the following: his identity an then be used to dedue the following: EW EW EW EW EV (1 ( θ θ Finally, we an show that the derivative with respet to gamma of the equivalent variation formula is less than the derivative of the equivalent variation for the high-inome region: EV θ (1 (1 Re-grouping and replaing θ with its value, the inequality is equivalent to the following: ( (1 ( ( whih is exatly the inequality we started with. In the examples till now there was a lose link between the equivalent variation of welfare gains or losses and hanges in real inome in both ountries. In ase of migration that link is not straightforward. hanges in real inome are no longer a good approximation of welfare gains. o illustrate this, assume migration from ountry to ountry inreases the labor fore in by μ perent, and new migrants

11 obtain the level of produtivity in the host ountry. s in the initial situation there is no trade and per apita prodution and onsumption of natives in both ountries remain the same. owever, total prodution and onsumption in the ountries do hange: EW EW EW EW EW s EW s (1 + μ (1 (1 + μ ( μ (1 ( μ EW s / ( + μ EW / ( μ s 10 (1 (1 + μ s in both ountries pries do not hange, real inome in ountry has inreased by ountry has delined by μ. his suggests a global inome gain of μ 10 μ and in 9. owever, this inrease overstates the global welfare gains and it is not a good approximation of EV. he problem is that aording to the real inome definition on a ountry basis, pries have not hanged. nd that is no longer an appropriate desription of pries relevant for individuals. It is still true for those who stay in the same ountry, but it is not true for new migrants. his is important beause all the welfare gains arue to new migrants in this example. For the migrants, inome has inreased tenfold after migration, but pries have inreased too. osts of living in the host ountry are signifiantly higher than in the ountry of origin as pries of non-tradable goods and servies are ten times as high. For the migrant utility before and after migration is U U OD M EW M Y Y / / Y ( Y / ( Y he money metri M of utility level U M at pries original ( and is M M ( U M,, θ ( U M + U M nd the EV-measure of welfare inrease for all migrants together is: U M EV EW OD OD OD OD OD [ M ( U,, M ( U,, ] [ U ew M U Old M ] μ [ Y Y ] μ μ μ [ 1 1] - 8 -

12 his is also the global welfare gain, whih is learly different from the gain in global real inome alulated in the standard way, i.e. the following ondition holds: 1 problem in the appliation of the formula above in applied general equilibrium (GE models is that those models normally do not ontain information of prie levels, nor about relative pries of non-tradable goods aross ountries. hese models initialize all pries at 1 by impliit hoie of the unit of measurement. his problem an be solved by deflating the new inome of migrants by a orretion fator, whih ompares prie levels aross ountries he deflated new inome ( 10 an then be ompared with the old inome (1 as both inomes are expressed in the same (base pries. he inrease in real inome after this orretion is again a good approximation of the EV of welfare gains by migrants and thus of the global welfare gains. Interestingly, inome sent bak home by migrants in the form of new remittanes should not be adjusted. hose who remain at home do not experiene a hange in prie. he impliation is that global welfare inreases if migrants from low-inome ountries send money bak to their ountry of origin. his seems ounterintuitive, beause aid flows or other transfers from rih ountries to poor ountries do not hange global welfare. he reason is not that we onsider remittanes different from aid flows or other transfers. he reason is that we have to treat migrants onsumption in host ountries differently. Before we apply the orretion in an empirial model, it has to be emphasized that the aggregation of welfare gains and losses is not done on the basis of pries. ggregation is done on the basis of market pries (the base pries before the shok. We only use a orretion funtion to transform urrent pries that migrants fae bak into original market pries they were austomed to in their home ountry. If welfare aggregation were to be done in pries, the whole analysis would be different. In that ase for example aid flow would inrease global welfare. By shifting resoures in donor ountries towards tradable goods and shifting resoures in the other diretion in the aid-reeiving ountries, prodution of non-tradable goods an inrease without a deline in the prodution of tradable goods

13 + + ( + (1 ( In analysis that would show up as an inrease in global real inome. In our analysis, real inome does not hange globally. But even in that ase spending by new migrants in host ountries have to be orreted for international prie differenes. 3. ppliation using the World Bank s IKGE Model hanges in nominal inome he model desribed above has many simplifying assumptions ompared to a standard GE model. mong other things, it exludes apital inome, it assumes same onsumption patterns, no hange in pries, and assumes goods are either perfetly homogeneous or perfetly non-traded. t the same time, as mentioned above, most GE models are alibrated to unitary pries in some base year aross all regions and therefore it is typially not possible to do the standard EV alulation, i.e. the relative value of ahieving the new utility level at the new (host pries ompared to the old (home pries. Instead, we use a shortut. We adjust the real onsumption of the new migrant in the host ountry by the relative exhange rate between the host and home ountries. able 1 desribes the proedure. he first panel of able 1 shows the hange in nominal after-tax inome for four groups of households natives in high-inome ountries (inluding migrants from other high-inome ountries, existing migrants in high-inome ountries, all remaining residents in developing ountries, and new migrants. Inome by soure is broken into three omponents apital and labor (after-tax inome and net remittanes. For natives in rih ountries, nominal apital inome inreases by $319 billion, offset by a deline in wage inome of $162 billion for a net inome gain of $158 billion. 1 For existing migrants, inome dereases by $100 billion omposed of a deline in wage inome of $125 billion, but offset by a deline in outward remittanes of $26 billion. For developing ountry residents, labor inome inreases by $170 billion, offset by a deline of $16 billion in apital inome and augmented by a net inrease in remittanes of $99 billion. otal nominal inome gain is $253 billion. For new migrants assuming they had no apital inome in their home ountry nominal inome inreases by $549 billion, of whih $126 billion is remitted bak for a total net inome gain of 1 here is a minor hange in remittanes for migrants from other high-inome ountries

14 $423 billion. he average per-worker after tax wage inreases from $4,900 to $43,000 (in he average after-tax wage in high-inome ountries for natives is $63,000. [he averages inlude both skilled and unskilled workers.] Globally, this leaves an inrease in nominal after-tax household inome of $735 billion. his amount is spent on the private onsumption of goods and servies and savings. Government expenditures are linked to GD. s GD rises in these simulations, government expenditures also rise. he nominal inrease in high-inome ountries is $88 billion, and $58 billion in developing ountries. ssuming government expenditures are alloated on a per apita basis, nominal expenditure on natives in high-inome ountries delines by $60 billion and by $4 billion for existing migrants. On the other hand, expenditure for new migrants inreases by $152 billion and unlike the expenditures for natives, it exatly mathes their tax payments, i.e. the shok is revenue neutral vis-à-vis the new migrants. he per apita numbers may be of interest. In the baseline, residents in high-inome ountries reeive an average of $10,480 in government goods and servies (exluding transfer payments. In developing ountries, the average per apita expenditure is $413, though $626 for the new migrants on average. dding the hange in personal onsumption expenditure and publi expenditure leads to a global nominal gain of $882 billion. he net nominal gain of $158 billion for rih ountry residents is offset by a transfer in publi expenditure of $60 billion to the new migrants for a total net gain of $98 billion. In developing ountries, the gain of $253 billion is augmented by an inrease in government expenditure of $58 billion, for a total of $312, of whih nearly $100 billion omes from an inrease in net remittanes. he new migrants get the bulk of the nominal gains. o their $423 billion nominal inome gain (after remittanes is added the $152 billion inrease in their onsumption of publi goods and servies. hanges in real inome he seond panel of table 1 shows the gains adjusted by hanges in pries. For residents in high-inome ountries, this reflets the nominal gains adjusted by the hange in the onsumer prie index (I for private expenditures and by the government prie deflator (GI for publi expenditures. he I barely hanges. hough the prie of non-tradables delines, this is offset by higher pries for manufaturing and apital-intensive goods and for imports. he GI on the other hand delines by 0.6 perent beause 2 he average wage of the new migrant is relatively high (ompared to the developing ountry average beause of the omposition of the new migrants that reflets a relatively high proportion from middle-inome ountries

15 government expenditures on goods and servies are onentrated in non-tradables. Similar adjustments are made for residents in developing ountries. In developing ountries the I inreases by 0.7 perent and the GI by 1.7 perent. he relatively high-loss of high-skilled workers is refleted in a higher ost of providing publi goods and servies. For the new migrants, their original basket of goods and servies is adjusted for the I in the home ountry. he new basket is adjusted for the I in the host ountry. Remittanes, on the other hand, are adjusted by the I in the home ountry, sine this is where they will be spent. In aggregate, the prie adjustment lowers the nominal gain in private expenditure from $735 billion to $554 billion. he largest adjustment (in perentage terms ours in the developing ountries due to the I adjustment. he adjustments in government expenditures are even more important. hese adjustments reverse the deline in rih ountries to near zero ompared to a loss of $64 billion in nominal terms. On the other hand, the rise in the ost in developing ountries lowers the nominal inrease from $58 billion to only $12 billion. In aggregate, the global gains amount to $674 billion in real terms, of whih $481 billion arues to the new migrants $372 billion in the form of higher wage inome after remittanes and $109 billion in the form of inreased expenditure on publi goods and servies. 3 adjustments he final adjustment is made to the purhasing power of the rise in real inome for new migrants for both private and publi expenditure on goods and servies. his is shown in the third panel of table 1. Only two numbers are different. he utility value of private expenditure of new migrants is adjusted for the relative purhasing power of the higher labor inome, i.e. the adjustment onerns the relative hange in the ost of living in the new host ountry relative to the home ountry. hus the real gain of $372 billion, drops to $126 billion. Similarly, publi servies are adjusted from a gain of $109 billion to $36 billion. he total net gain for new migrants then beomes only $162 billion and the net global gain is $356 billion. he implied exhange rate is 2.3 that is somewhat lower than the developing ountry average (3.2 on a GD-weighted basis. he main reason is that migrants on average are oming from the higher-inome developing ountries where the exhange rates are typially loser to the market exhange rates. 3 hough the GI delines in high-inome ountries this is only true for existing residents. For new migrants, there is also an adjustment for the differenes in the GI between the home and host ountries. he GI grows relatively more rapidly in rih ountries in the baseline than in developing ountries

16 able 1: Soure of gains in standard simulation $billion in 2025 apital abor Remittanes Subtotal Govt. otal ominal gains igh-inome atives MY migrants otal Developing ew migrants Global Real gains igh-inome atives MY migrants otal Developing ew migrants Global Real gains adjusted igh-inome atives MY migrants otal Developing ew migrants Global onlusions In this paper we disussed two sets of issues that arise when one alulates global gains from migration. Some aspets of the disussion are speifi to migration, other parts of the argument are also relevant for the analysis of other poliy shoks. he first set of issues is related to how the gains of speifi groups

17 should be evaluated. he seond set of issues is related to how the gains should be aggregated over groups and ountries. Evaluation of the welfare gains of individual groups. In standard appliations of general equilibrium (GE models, the welfare impats of speifi groups are evaluated using a onept from welfare theory alled equivalent variation (EV. he onept is relatively straightforward. Welfare hanges as a result of hanges in nominal inome and hanges in pries. EV alulations summarize this welfare hange in terms of an equivalent hange in inome alone, showing by how muh inome at original pries would have to hange to ahieve the same hange in welfare as observed in a simulation. 4 For most households, the standard notion of the hange in real inome, that is, the differene in nominal inome adjusted by the hange in the I, is a good approximation of EV. 5 his is not the ase for new migrants, however. here is no standard prie index that an be used as a deflator for the hange in the nominal gains for the new migrants, sine the pries they fae in their new host ountry have no linkages to the pries they paid in their home ountries. GE and maro models typially alibrate base-year pries in eah region to one (or unit value by hoosing orresponding volume units. 6 his approah does not allow one to take into aount the prie inreases that new migrants fae as a result of their migration. In the migration simulations desribed above, the maro exhange rate (as an approximation of the rise in pries faed by migrants from developing to industrial ountries has been used to adjust the gains to the migrants although this is just an approximation of the true welfare gains. Beause of the ost-ofliving adjustment to the welfare gain of new migrants, the real gain reported is no longer equal to real inome gains of ountries and real output gains measured using national aounting standards. owever, the standard real inome measure is still a good approximation of the welfare gains for the other households in the model. o the extent that new migrants remit part of their inome to their ountry of origin and that inome is spent in that ountry of origin, the inrease in the ost of living that new migrants fae is not relevant. herefore, the EV measure of remittanes is larger than the same nominal inome spent by the new 4 One of the advantages of the EV measure is that it transforms the ordinal onept of welfare into a ardinal onept of inome. While it is impossible to measure how muh one welfare level differs from another (one an only onlude that one level is preferred to another, the orresponding inrease in inome an be measured, and the size of the inrease has a lear meaning. 5 For example, in trade-reform senarios, the hange in the prie index is a relatively good approximation of the welfare impat, sine the new prie is approximately the old prie less the tariff. 6 here are exeptions. For example, in the ase of limate-hange models, it is neessary to know the relative pries of the different fuels to aurately determine the arbon tax

18 migrant in the host ountry. his differene illustrates the inentive for new migrants to remit inome home. ggregation of gains. he seond issue relates to the interpretation of the global gains. ypially, to derive aggregate or global gains, EV (expressed in a ommon urreny, typially the U.S. dollar is summed aross all households. For individual persons or homogeneous groups this EV aggregation, expressed as a perentage of original inome, is a good approximation of the hange in welfare (or more preisely, it is a good indiation of the hange in welfare. 7 owever, no lear link exists between global welfare and the aggregation of EV aross heterogeneous groups, beause we do not know how to weigh individual welfare aross heterogeneous groups (a partiularly diffiult issue in aggregating aross ountries at very different stages of development, as is done here. For example, while most groups gain from migration in the senario disussed in the text, some lose. he fat that the hange in global welfare (expressed as the aggregation of EV aross groups is positive does not mean that the welfare gains of the winners are onsidered more important than the welfare losses of the losers. hus, global gains as expressed in aggregate EV should not be interpreted as a value judgment on how to weigh individual or loal welfare gains. he aggregation of EV aross groups does, however, have a useful interpretation, whih is linked to the notion of ompensation and areto optimality. s long as the global gains are positive using the standard pratie of adding up EVs aross households then it is possible through redistribution to ompensate households that lose so that no one is worse off relative to the baseline senario, while some are better off. In that sense the global gain an be ompared with an equal rise in global output plus redistribution. n alternative approah to alulating global gains would be to add up hanges in inome measured in terms. he rationale for that alternative is that beause pries of non-traded goods are lower in developing ountries, the addition of a dollar to a developing ountry would enable the purhase of a larger amount of goods and servies than in an industrial ountry. In that ase, both base inome and gains for new migrants and for those who remain in developing ountries would be roughly three times as large as reported here. his is true for all gains, whether they ome from migration itself, from remittanes, or from hanges in wages and pries in developing ountries. s a result, the share of those who (originally live in developing ountries in global aggregates would inrease in the measurement of both global inome levels and global welfare gains. owever, the perentage inrease in inome for developing ountries would not be affeted. 7 he size of the hange in individual welfare is undetermined, sine welfare is an ordinal onept

19 Obviously, global inome and global gains would also be larger if expressed in terms. s the perentage inrease in welfare for people (originally living in developing ountries is larger than the perentage inrease for those living in high-inome ountries, a swith to measures would also inrease the global gains as a perentage of global inome. If in the migration senario presented here the gains are -adjusted, the global gains would amount to 0.9 perent of global inome in the baseline, instead of 0.6 perent using the EV aggregation. his illustrates that people (originally living in developing ountries are the biggest gainers of migration in perentage terms