sian Growth and frian evelopment Maros Chamon Mihael Kremer 1 January 2006 Sine World War II, integration with the world eonomy has arguably been the hief route from poverty to wealth. Japan initially exported heap goods and later moved on to more tehnologially sophistiated produts. When Japan beame rih, Korea, Taiwan Provine of China, Hong Kong SR, and Singapore replaed Japan as low wage exporters, and when these ountries moved on to more sophistiated produts, Thailand and Malaysia filled their nihe. More reently, China has beome an important exporter of manufatured goods and India is inreasingly moving into servies exports. o mainland sub-saharan frian ountry has experiened this type of transformation. Even ountries that have undergone major eonomi reforms seem far from takeoff. Foreasts for fria based on extrapolation of its historial experiene tend to be bleak. We onsider whether it is possible to onstrut a model, onsistent with the data, whih supports a more optimisti view. Motivated by the sian experiene, the model assumes ountries an potentially undergo rapid eonomi transformation only if they integrate into the world eonomy by produing non-traditional exports. For the purposes of onstruting a long-run model of the world eonomy, we are agnosti on whether exports matter due to tehnologial learning by doing spillovers, politial eonomy onsiderations, or other fators. s eah developing ountry transforms and beomes advaned, it further improves trade opportunities for the remaining developing ountries. For example, if China beomes rih, a billion more people will live in ountries that import toys and a billion fewer will live in ountries that export them. 1 Chamon: Researh epartment, International Monetary Fund, e-mail: mhamon@imf.org. Kremer: Gates Professor of eveloping Soieties, epartment of Eonomis, Harvard University, Senior Fellow, Brookings Institution e-mail: mkremer@fas.harvard.edu. The views expressed in this paper are those of the authors and should not be attributed to the IMF its Exeutive Board, or
Our approah is similar to that of Robert E. Luas (2000) in that we assume that growth prospets improve with the state of the world eonomy, potentially generating aelerating world growth. However, unlike Luas (2000), we allow for differential population growth between rih and poor ountries. The steady-state proportion of the world population living in advaned ountries depends on the rate at whih developing ountries transform into advaned ones and on the magnitude of population growth differentials between these two groups of ountries. The eonomi transformation proess will overome the demographi trend, leading to a prosperous steady state, only if the initial share of world population is above a threshold. simple alibration using historial data suggests that the long-run prospets for lagging developing regions may hinge on a rae between eonomi growth in China and India and population growth in the lagging regions, partiularly fria. If the former wins, we may eventually observe aelerating global growth. This suggests some aution should be used when interpreting empirial studies on growth determinants, as these results may not be stable over time. Country harateristis that lead to poor performanes today may well allow for rapid growth in the future if and when the world eonomy reahes a suffiiently advaned stage. Our model also suggests a queuing effet, where the order in whih ountries are absorbed into the world eonomy is determined by the quality of their poliies. Eonomi reforms in one ountry may potentially have a large impat on growth if they move the ountry to the front of the queue. But similar reforms in all ountries may have a muh smaller impat on world growth. I. The Model Consider a model with two prodution tehnologies: traditional and modern. Labor is the only input and is inelastially supplied. In the traditional tehnology, eah unit of labor produes one unit of final onsumption. The modern tehnology ombines a simple intermediate input L and a omplex intermediate input H to produe the onsumption good. Trade allows these tasks to be performed in different ountries. For its management. We are grateful to avid Weil, for suggestions. Key words: Population, growth; eonomi development; international trade. JEL lassifiation: J11; F43; O41. 1
example, the simple input ould be stithing shoes and the omplex input ould be designing and marketing the shoes, or the simple input ould be automobile assembly and the ompliated input ould be manufaturing automobile omponents. There are many small ountries of two types: advaned and developing. Only advaned ountries an produe intermediate good H. Intermediate good L an be produed anywhere, but prodution in a developing ountry involves an additional ost of δ units of the final good. This an be thought of as a transport ost or as a ost aused by poliy distortions. Thus: Y traditional Y modern H L = n traditional, α 1 α = H L nh, if advaned = 0 if developing nl, if advaned = nl, + δ Y if developing where > 2(1 + δ ), 0 < α < 1, n traditional,, nh, and n L. are the number of workers in ountry engaged in the traditional prodution and the prodution of intermediate goods H and L respetively. Eah period, developing ountries beome advaned with probability p times the share of their population in modern prodution. This is not internalized by produers nor workers. We assume population growth in advaned ountries, γ, is less than that in developing ountries, γ, as has been the ase empirially sine the early 20th Century. Eah period, advaned ountries allow immigration from developing ountries equivalent to a proportion i of the advaned ountries population. I. Evolution of the World Population Modern setor firms hire developing ountry workers until the marginal revenue produt equals their wage (one if the outside option is the traditional setor) plus the transation ost. In a ompetitive equilibrium, advaned ountries will only demand L from a developing ountry one all the developing ountries with lower transation osts 2
have already joined the modern setor. epending on how large these differenes in ost are, the modern setor may never reah the highest ost ountries. We fous on a favorable senario for lagging developing regions, where the differene in transation osts aross ountries are small: δ δ. We later disuss an alternative and less favorable senario. eveloping ountry modern setor employment is therefore M 1/ α = min(( (1 α ) /(1 + δ )), ), where and developing ountries' population. We onsider the ase where the initial suffiient to absorb all into the global eonomy, so:.. = γ = γ This implies that if: γ γ (1) 1, 1/ α p( (1 α) /(1 + δ )) + i then.. + p( (1 α) /(1 + δ )) 1/ α 1/ α ( p( (1 α) /(1 + δ )) + i). denote the advaned and + i, is not / < / and the proportion of the world population in advaned ountries delines indefinitely, onverging to zero. If population growth rates are similar in advaned and developing ountries, / will inrease over time. If not, the / ratio will only inrease if it's starting value is above a threshold equal to the right-hand-side of (1). If that is the ase, the transformation and migration proesses dominate the opposing demographi trend. One all workers are in the modern setor, wages in developing ountries are no longer pinned down by traditional setor produtivity and will grow with /. epending on parameter values, / will either grow without bounds or onverge to a steady-state ratio. 2 The model implies that differenes in transation osts between ountries have strong impliations for whih developing ountries will integrate into the global eonomy first. Even when the differenes are arbitrarily small, the ordinal rank of these osts will 3
plae ountries in a development queue. n individual ountry benefits from poliy improvements that redue its transation osts, if that moves it to the front of the queue. However, this omes at the expense of other developing ountries. The limited ability of advaned eonomies to absorb the labor in the developing world means that poliy improvements have smaller impats on world growth than on individual ountry growth. That is, individual ountry growth an be driven by ordinal improvements while world growth depends on the atual magnitude of the improvements. This an help explain growth failing to pik-up in the developing world as a whole despite signifiant improvements in poliy over the last deades (as doumented in William R. Easterly 2001). The model also suggests a non-linear impat of poliy reform, with growth potentially responding dramatially if the reforms move a ountry to the front of the queue but not otherwise. frian ountries may have limited opportunities for integration over the short- and medium-term if they are (and remain) behind sian ountries on this development queue. If transation ost differenes aross ountries are signifiant, the proess of global integration will slow down as it reahes inreasingly more ostly developing ountries. In the extreme ase in whih a subset of ountries have poliies bad enough α (1 α ) that the resulting transation ost is δ > α (1 α) 1, these ountries will never be integrated into the global eonomy, beause the advaned eonomies would rather produe L themselves than transat with these ountries. If that is the ase, in the absene of migration, the share of the world population living in these ountries would grow indefinitely sine δ > δ. However, migration from these ountries to advaned ones an ompensate for the natural population growth differential if / is suffiiently high. Provided / > γ /( γ i), the global eonomy will still onverge to widespread prosperity. + II. Simple Calibration We alibrate the model using population and GP data from ngus Maddison (2003) for 1820 2001 and population and international migration data and projetions 2 See Maros. Chamon and Mihael R. Kremer (2006) for details. 4
from the United ations (U..) Population ivision for 1950 2050. We lassify eonomies as developing or advaned based on whether their GP per apita, measured in 1990 International Geary-Khamis dollar terms, was higher than one third of that of the leading ountry, defined as the United Kingdom for 1820 1900, and the United States afterwards. Sine the model assumes one-way transitions from developing to advaned status, we fous only on eonomies that permanently ross that threshold (exeptions are made for temporary delines). See Chamon and Kremer (2006) for details. Figure 1 shows the relative population in the advaned world, /, inreased throughout the 19th entury beause population growth in advane ountries exeeded that in developing ountries. In the early 20th entury, population growth began delining in advaned ountries and inreased substantially in developing ones. Following this demographi reversal, / gradually delined through the 20th Century, with blips when large ountries transitioned from developing to advaned eonomy status (notably Japan in 1932). ote that of advaned eonomies. We estimate / delined despite a substantial inrease in the number 1/ α p ( (1 α) /(1 + δ )), the arrival rate of transition from developing to advaned eonomy status, as 0.40% over the last 50 years. This is the average ratio of the population in the eonomies that beame advaned in a given year to the total advaned eonomy population. uring 2000 05, natural population growth rates were 0.29% in advaned eonomies and 1.38% in developing ones, and net migration into advaned eonomies was 0.31% of their population. The U.. Population ivision projets natural population growth rates to ontinue to deline in both advaned and developing ountries, with their gap narrowing but remaining substantial. atural population growth in fria is projeted to remain substantially above the developing world s average (Figure 2). Under the parameter values above, / would be below the ritial threshold and the proportion of the world population living in developing ountries would grow indefinitely, just as it has grown sine the early 20th entury when modern mediine reahed the poor world. Under the demographi hanges projeted by the U.., the world will eventually onverge to prosperity but the proess would be extremely slow. o 5
substantial improvements would take plae within a entury. s shown in Figure 3, the long-run dynamis ould substantially improve through halving natural population growth disparities, doubling immigration or doubling the transformation rate. The transformation rate ould potentially hange due to improvements in ommuniation and transportation or poliy improvements. The model and this alibration exerise have onsidered the world eonomy as a large olletion of small ountries, whose evolution an be approximated by a smooth and deterministi proess. In pratie, the transformation of large ountries ould move sizable shares of the world population from poverty to prosperity, adding stohastiity to the world eonomy. Figure 3 also shows the effet of China and of China and India instantaneously beoming advaned eonomies. This leads to a large aeleration in the rate at whih other eonomies develop. Thus, even if China and India are at the front of the development queue, that ould atually benefit other developing ountries in the longrun provided these giants transform suffiiently rapidly. III. Conlusion model in whih development omes through non-traditional exports suggests that the prospets for eonomi transformation in developing ountries depend on the amount of labor that an be absorbed by the advaned eonomies. fria s nontraditional export opportunities may remain limited as long as labor remains heap in China and India. But one China and India develop, there will be a very strong demand for the labor in the remaining developing ountries, potentially leading to aelerating growth over time. The model also suggests that the development of these two large ountries ould help other developing ountries in the short- and medium-term through terms of trade effets. Our results suggest one should be areful when extrapolating the existing empirial evidene on eonomi growth into the future. Over short horizons, these results are likely to exaggerate the developing world-wide benefits of poliy improvements, as some of the observed growth may ome at the expense of other developing ountries if the queuing effets desribed are at play. Over longer horizons, these empirial results 6
are likely to be over-pessimisti, as ountry harateristis that lead to low growth today may allow for high growth if and when the global eonomy reahes a suffiiently advaned stage. The old onventional wisdom was that population growth in developing ountries was a major problem. The new onventional wisdom is that population growth is rapidly delining and hene not an obstale to development. Our results highlight the importane of relative population growth between rih and poor ountries. Beause population growth has been delining in the rih world along with the poor world, this population growth gap is relatively persistent. Its worth noting that in our model, population growth in developing ountries will reate negative externalities on other ountries, while population growth and open immigration poliies in advaned ountries will reate positive externalities for the rest of the world. The model suggests that the future of the world eonomy may well be deided by a rae between sian eonomi growth and frian population growth. Referenes Chamon, Maros. and Mihael R. Kremer (2006). "Eonomi transformation, population growth and the long-run world inome distribution.", IMF Working Paper (Washington: International Monetary Fund, forthoming). Easterly, William R. (2001). The Lost eades: eveloping Countries Stagnation in Spite of Poliy Reform 1980-1998. Journal of Eonomi Growth, 6(2): 135-157. Luas, Robert E. (2000). ''Some Maroeonomis for the 21st Century.'' Journal of Eonomi Perspetives, 14: 159-168. Maddison, ngus (2003). The World Eonomy: Historial Statistis. Paris and Washington,.C.: Organization for Eonomi Cooperation and evelopment. 7
Figure 1: Ratio of World Population in dvaned and eveloping Eonomies and umber of dvaned Eonomies. Pop. dvaned/pop. eveloping.15.2.25.3 Pop. dvaned/pop. eveloping umber of dvaned Eonomies 15 20 25 30 35 umber of dvaned Eonomies 1800 1850 1900 1950 2000 Year otes: ata from Maddison (2003). Missing observations log-linearly interpolated. 8
Figure 2: atural Population Growth and et Migration from 1950 to 2005 and Projetions for 2005 to 2050. % 0 1 2 3 atural Pop. Growth fria atural Pop. Growth eveloping atural Pop. Growth advaned et Immigration Rate to dvaned 1950 1975 2000 2025 2050 Year otes: ata from the United ations Population ivision. Plot indiates average migration rate and the (geometri) average growth rate over the 5 year period ending in that year. Solid lines indiate atual values and dashed lines indiate projetions. 9
Figure 3: Evolution of the Ratio of World Population in dvaned and eveloping Eonomies, Exluding China and India: Pop. dvaned/pop. eveloping Exluding China and India.2.3.4.5.6 China and India Beome dvaned Baseline Senario atural Pop. Growth Gap Halves Transition Rate oubles Immigration Rate oubles China Beomes dvaned Projeted emographi Changes 2000 2020 2040 2060 2080 2100 Year otes: The baseline senario orresponds to the parameter values as of 2000 2005, and the historial rate of transition from 1952 2001. 10