94196 Ea S tA pe s s ci ia ResearchDigest al an Is d World Bank su Pa e ci on fi c VOLUME 9 L NUMBER 2 L WINTER 2015 Facilitating Labor Migration from the Philippines Labor migration brings big gains •  Information about how to apply— IN THIS ISSUE in income. Can source countries providing details on typical overseas Facilitating Labor Migration from the increase such migration through costs, steps needed to apply for work Philippines … page 1 abroad, enrollment on an overseas policies implemented unilaterally? Can developing countries increase labor job-finding website (designed as part emigration through unilateral policies? M of the project), and ways to avoid ille- oving from a developing to gal recruitment. Incentives and Teacher Effort … page 2 a developed country results •  Information about how to finance An analysis of school survey data from Lao in immediate large gains in migration—providing details on typical PDR looks at what kinds of incentives can income for the migrants—gains far placement fees for work abroad and a increase teacher effort exceeding those from any other devel- list of Manila-based financial compa- opment policy intervention. Yet very nies that offer loans for such fees. Electoral Accountability and Local Public Spending in Indonesia … page 3 little development policy (or research) •  Job-matching and website assis- is devoted to trying to allow more peo- tance—helping people enroll on a jobs Indonesia’s phased introduction of direct ple to take advantage of this oppor- website, developed and set up in col- elections of local leaders allowed analysis of how electoral accountability affects public tunity. We swoon over interventions laboration with several Manila-based spending that raise poor people’s incomes by recruitment agencies, where applicants the equivalent of a few cents a day, yet could match with recruiters. Why Don’t Poor Countries Do ignore the possibility that they would •  Passport assistance—fully subsidiz- R&D? … page 4 be far better off if they could move ing the costs of obtaining a passport. Poor countries invest far less in research and somewhere else. (The project learned that recruitment development as a share of GDP than rich The Philippines is well known for its agencies prioritize applicants with ones do. That may be a rational approach bilateral efforts to facilitate migration passports and that getting a passport through arrangements with different is surprisingly time-consuming.) How Effective Are Efforts to Raise Voluntary countries. These efforts, coupled with a The work was done in Sorsogon, Enrollment in Health Insurance? … page 5 well-regulated private recruiting indus- a province that sends relatively few In the push toward universal health coverage, try, have resulted in an annual deploy- migrants overseas. The main sample how much can measures to raise voluntary ment of about 2 million Filipinos to consisted of 4,151 individuals selected health insurance enrollment help? countries around the world. Yet even through door-to-door surveying in How Insecure Property Rights Affect in the Philippines—with a per capita randomly selected barangays. These Migration in China … page 6 income of only $2,600 ($4,400 in PPP individuals were randomly placed in Institutions help shape the decision to terms)—more than 95 percent of the treatment groups receiving different migrate from rural to urban areas in China. labor force is not working abroad. interventions, including combinations One of them is land tenure security How effective are unilateral efforts of interventions, with the group receiv- by source countries to facilitate more ing all the interventions called the “full The Challenges of Public Service Delivery in international migration? A recent assistance” group. the Pacific Islands … page 7 paper by Beam, Yang, and McKenzie The key results are these: A new paper contests the conventional view reports on a project that investigated •  Demand for migration is much that public sectors in the Pacific island this question through several years of lower than income differences would suggest. countries are too big experiments in the Philippines. The Only 34 percent of the sample said at project tested several interventions de- baseline that they were interested in signed to overcome different potential constraints to migration: (continued on page 8) 2 World Bank ResearchDigest Incentives and Teacher Effort School survey data from Lao PDR of them for no fee. Teachers generally with a 0.12 percent increase in the show that nonmonetary incentives spend an average of 13 hours a week probability that teachers offer tutoring, can be effective in motivating preparing for class and grading home- while the freedom to develop teach- work. Nevertheless, teachers report ing materials correlates with teachers teachers to increase their effort frequent delays in salary payments. working three additional hours a week. T They report a delay of 1.8 months on Incentives such as teachers having eachers and their classroom average, with a maximum of up to 7 autonomy over the teaching method performance affect the ability of months. or the principal having the power to any education system to produce The school survey collected richer dismiss teachers lead teachers to learning results, and incentives fea- information related to monetary and substitute effort away from the less ture prominently in this process. For nonmonetary incentives. The authors visible activity of preparing for classes example, even when teachers are well use reported delays in the payment of and toward the more visible activity of trained, they may not be motivated individual teacher salaries to capture after-school tutoring. to do their best because good per- monetary incentives, and the extent One limitation of this study is that formance is not aptly rewarded, with of teacher autonomy, parent-teacher the school survey did not collect data salaries being determined primarily by association (PTA) activities, and a on student performance such as test seniority, or because performance is school principal’s authority to measure scores, so the impact of incentives and neither monitored nor measured. nonmonetary incentives. Interestingly, teacher effort on student achievement To improve performance, some these incentives represent the cur- cannot be investigated. But related countries have linked at least part of rent practice in Lao PDR’s education studies point to the general positive teacher pay to performance criteria system and are relatively amenable to influence of increased teacher effort such as student test performance. policy influence in this country as well (as measured by lower absenteeism, Opponents of this approach argue, as in other developing countries, but for example) on student learning, however, that the work of teachers is most have not been analyzed in previ- which suggests that a higher prob- multidimensional, with only some of ous studies. ability of teachers providing tutoring its aspects measured by student test The authors also investigate novel or longer after-school work hours are scores, and that linking pay to student measures of teacher effort that include likely to result in nonnegligible and performance would lead to teachers the number of after-school hours a beneficial effects on students. teaching mainly to the test. Moreover, teacher works, whether or not a teach- Promising avenues for future re- research suggests that nonmonetary er offers after-school tutoring, whether search include more investigation and implicit incentives, such as work or not after-school tutoring is for pay, into the strategic behavior of teachers conditions and peer pressure, may be and how many students a teacher tu- under different incentive mechanisms. sufficient or even more powerful in tors after school. The first two main Moreover, because of measurement is- raising teacher effort. outcomes—after-school work hours sues related to self-reported effort, di- The Lao People’s Democratic and after-school tutoring activities— rect observation methods akin to those Republic offers a particularly interest- are jointly estimated with a new simul- used in time allocation studies— ing case for analysis. This country has taneous ordinary least squares–probit methods that are admittedly costly to witnessed steady economic growth model with school random effects, an implement—would help strengthen rates averaging 7 percent in recent approach that can provide more effi- the quality of future studies. years, but it remains one of the poor- cient estimates than a single-equation est countries in East Asia, with limited estimation approach. Other economet- quantitative evidence on its education ric models, such as multinomial logit system. A recent paper by Dang and and negative binomial models, are King investigates the link between also used. teacher incentives and effort in Lao The authors find that both mon- PDR using a nationally representative etary and nonmonetary incentives primary school survey fielded by one correlate with increased teacher effort of the authors. in after-school work hours and after- Basic summary statistics reveal school tutoring activities. In particular, interesting patterns in the data. For one month of salary delay is associ- example, teacher morale is higher and ated with a 40 percent decrease in the Hai-Anh H. Dang and Elizabeth M. King. 2013. absenteeism lower than in other coun- odds of teachers providing tutoring “Incentives and Teacher Effort: Further Evidence tries at higher income levels. Thirteen lessons free of charge and a 30 percent from a Developing Country.” Policy Research percent of teachers offer tutoring les- decrease in the number of their tutees. Working Paper 6694, World Bank, Washington, sons to their students, more than half The existence of a PTA is associated DC. World Bank Research Digest 3 Electoral Accountability and Local Public Spending in Indonesia A shift to direct elections of local on human development outcomes and conclusion reinforced by the finding leaders in Indonesia was followed the pattern of public spending and rev- that the increase in district spending by increases in local spending on enue generation at the district level. in health was financed by a signifi- The paper assembles a large data cant decrease in “other” discretionary health. Will they be sustained? set, compiled from multiple sources spending. T linked at the district level, such as One important question emerging he relationship between institu- district-level data on electoral infor- from the analysis that merits further tions of political accountability mation, key outcome measures at the exploration is whether the increases and government performance has district level constructed from the an- in district spending on health are long been a concern. There is growing nual national socioeconomic surveys sustained over time. These spending recognition that fiscal decentraliza- of households (SUSENAS), and district increases may well reflect the incentive tion alone will not lead to more effi- revenue and expenditure data for the of district heads to win popular sup- cient allocation of resources. Political years 2001–09. port for reelection. But once reelected institutions shape incentives, and Using the difference-in-differences for a second term, district heads face fiscal decentralization requires politi- estimator, the paper reveals that four term limits that may weaken their in- cal institutions that not only balance years after the switch to direct elec- centive to continue supporting an ex- accountability and representation at tions the reforms had had no effect on pansion of health programs. There is a the local level but also help align lo- human development outcomes. But potential countervailing force as well: cal political incentives with national the composition of district expendi- rising party loyalty of district heads interests. tures appears to change considerably. and the continued reputation of po- A recent paper by Skoufias, During the year of and sometimes litical parties with district voters may Narayan, Dasgupta, and Kaiser looks the year before the elections there is eventually become sufficiently strong at how institutional design for political a significant increase in the types of forces to counteract the individual in- incentives and electoral accountability district expenditures that could allow terests of district heads. affects public revenue and expendi- the incumbent district heads running The empirical evidence available, at ture choices at the district level in in the direct elections to “buy” voter least from the United States, suggests Indonesia. After the fall of Suharto, support—specifically, expenditures in that political parties are a weak coun- fiscal decentralization reforms as- personnel and “other” discretionary teracting force against the interests of signed local district governments categories. individual elected officials facing term wide-ranging responsibilities over The analysis also suggests that limits. Moreover, the prevailing percep- basic service delivery in health, educa- direct elections may have increased tions in Indonesia about the weakness tion, infrastructure, and general public the responsiveness of district govern- of party ideology, lack of party loyalty administration. In addition, changes ments to local needs at least in the among district heads, and general were introduced in the way local ex- area of health. Direct elections have voter mistrust of political parties sug- ecutive leaders (district heads) were been followed by a significant increase gest that the role of national political selected. in district expenditures in health—an parties as a disciplining device for dis- While fiscal decentralization was increase that may be due in part to trict heads—able to align local politi- implemented in a “big bang” in 2001, district heads using their discretion cal incentives with national interests the method for selecting district heads to provide local health insurance for and sustain policy changes over the was changed in a phased manner. the poor and near-poor. This finding is longer term—may take a long time to These officials were initially selected some of the first, though indirect, evi- materialize. by the local legislatures. But starting dence that political decentralization, in 2005 they were selected through following fiscal decentralization, pro- direct elections by citizens, with the vides incentives for local governments timing of the first election in each dis- to become more accountable to poor trict determined by whether the head citizens. selected by the previous system had The analysis finds that electoral served his or her full tenure. By June reforms did not lead to higher revenue 2005 direct elections had been held in generation from own sources and had a little more than a third of districts, no effect on the budget surplus of and by 2007 in around 62 percent. The districts with directly elected heads. Emmanuel Skoufias, Ambar Narayan, Basab paper takes advantage of the rare “nat- This suggests that the greater electoral Dasgupta, and Kai Kaiser. 2014. “Electoral Ac- ural experiment” provided by the stag- accountability for local governments countability and Local Government Spending in gered implementation of electoral re- was accompanied by a generally pru- Indonesia.” Policy Research Working Paper 6782, form in Indonesia to measure impacts dent approach to fiscal balances, a World Bank, Washington, DC. 4 World Bank ResearchDigest Why Don’t Poor Countries Do R&D? Poor countries invest much less in Yet poor countries invest far less in and the likelihood that it will be com- R&D than the high measured rates R&D as a share of GDP than rich coun- municated to the private sector fall of return suggest they should. But is tries do (figure 1). The two important among poorer countries, suggesting an exceptions are China and India (follow- important offset to the Schumpeterian this in fact rational? ing Israel, Finland, and the Republic benefit to lagging. E of Korea previously), whose dramatic The paper estimates the pattern stimates of the returns from re- trajectories suggest that they, in fact, of returns to R&D across the develop- search and development (R&D) think about it a lot. ment process as suggestive evidence for advanced countries are so In a recent paper, however, Maloney that such offsetting effects may be im- high, the literature has argued, that and Goñi argue that developing coun- portant. It constructs a global panel of they justify levels of investment that try governments and firms may well R&D expenditure across 40 years for 75 are multiples of those actually found. be rational: there may be countervail- countries, a broader sample of rich and The case may be even stronger for ing forces that prevent returns from poor countries than has been used be- poor countries, where a long literature continuing upward with distance from fore. Employing recent advances in in- argues that R&D is essential to the the technological frontier. In particular, strumental variable varying coefficient “absorptive” or “national learning” ca- the necessary complementarities to estimators to allow returns across all pacity required to exploit technologi- R&D spending are likely to diminish factors to vary across the development cal advance in the advanced countries. with distance from the frontier and space, it confirms the earlier findings In fact, earlier empirical research hence reduce the efficacy of a given of increasing returns with distance in OECD countries showed that the unit of R&D. Education, the quality of from the frontier. However, returns estimated returns to R&D rise with dis- scientific infrastructure, and the overall then begin to decline, following an tance from the technological frontier, functioning of the national innovation inverted U shape. For countries further increasingly reflecting the greater gains system—which, together, translate from the frontier than, for example, from catch-up afforded to follower spending into knowledge—lag in poor modern-day Mexico, the returns fall off countries. Extrapolating the estimates countries. Technological advances at and eventually become negative. These to even middle-income countries sug- the frontier may not easily translate results are robust to several different gests truly large potential returns, into advances for a developing country specifications. justifying a much larger R&D effort in because of different technologies pres- From a policy point of view, this developing countries than is found in ently applied or because of skills mis- suggests that focusing narrowly on the advanced ones. To reframe Robert match. To the degree that poorer coun- increasing R&D can be counterproduc- Lucas’s famous observation about tries have less sophisticated firms, or tive, diverting resources from more growth, confronted with the rates of simply less accumulated learning by mundane needs such as infrastructure return found in the literature, govern- doing, the translation of R&D into mar- or basic education with little gain. ments would find it hard to think of ket returns is weaker and the quality of The development of innovative capac- anything else. the feedback from the private to the re- ity requires an across-the-board im- search sector diminishes. Both the ef- provement in all the necessary factors fective output of R&D per dollar spent complementary to R&D. The question arises, then, why China and India are Figure 1. R&D Spending by Level of Development, Selected Economies investing so much in R&D if, for coun- R&D as % of GDP tries at their level of development, the 4.5 likely returns are small. A likely expla- 4.0 Finland nation is that they effectively import Japan the complementary factors in the form 3.5 Korea, Rep. of multinational corporations that do 3.0 much of the patentable research. The 2.5 degree to which the returns to this Predicted R&D “offshore” R&D accumulate to the host 2.0 Taiwan, China country remains to be researched. 1.5 China Spain 1.0 Brazil Turkey Hong Kong SAR, China India Malaysia 0.5 Mexico William F. Maloney and Edwin Goñi. 2014. 0 Colombia “Why Don’t Poor Countries Do R&D?” Policy 4 5 6 7 8 9 10 11 Research Working Paper 6811, World Bank, GDP per capita (log of constant US$) Washington, DC. World Bank Research Digest 5 How Effective Are Efforts to Raise Voluntary Enrollment in Health Insurance? An experiment in the Philippines where despite numerous initiatives the in April 2012 the enumerator could suggests that achieving universal informal sector remains the population help them complete their enrollment health coverage through voluntary group with the lowest coverage rate. form, deliver the form to the insurer, The main experiment and arrange for their enrollment will not be easy tested the effective- Much of the world identification card to be M ness of a combination sent in the mail. This uch of the world is in the of information and a is in the midst assistance “treatment” midst of a push toward uni- premium subsidy in worked: enrollment was versal health coverage. One of raising enrollment in of a push toward 3.4 percent among fami- the biggest challenges facing develop- the voluntary govern- universal health lies that did not receive ing countries is extending insurance ment-run Individual coverage to informal sector workers. Payer Program (IPP). coverage. One of the assistance from the enumerator but as high These workers are very mobile, and From a sample of 243 biggest challenges as 39.7 percent among since they are also typically self- municipalities, 179 were those that did. employed, they do not have employ- randomly assigned as facing developing The dramatic in- ers who can be required to deduct intervention sites and countries is crease in coverage in premium contributions from their 64 as controls. In early salaries. Since they are not officially 2011, 2,950 families extending insurance the second experiment points to the impor- poor or marginalized and because were interviewed, and coverage to informal tance of the transport they represent a sizable fraction of the unenrolled IPP-eligible and time costs associ- population, many countries are reluc- families in interven- sector workers ated with health insur- tant to fully subsidize their coverage, tion sites were given ance enrollment—costs though some have done so (Thailand an information kit and a 50 percent that were highlighted by one partici- is an example). premium subsidy voucher valid until pant who texted in reply to one of the Many countries have instead used the end of 2011. The randomization SMS reminders that she had missed partial subsidies to encourage enroll- process worked as planned: the control the boat and therefore could not get ment either in a new government-run and intervention sites were “balanced” to the provincial insurance office that scheme, as in China and Mexico, or at baseline—that is, they were similar day. This experiment is also sobering, in the government’s existing social in terms of variables likely to influ- however: even with all the assistance health insurance program, as in the ence IPP enrollment. By January 2012, provided to the treatment group, 60 Philippines and Vietnam. Some coun- 9.9 percent of eligible individuals in percent chose not to enroll. With the tries (including China and Mexico) the control sites had enrolled, com- costs of providing such “hand-holding” have achieved high coverage levels, pared with 14.9 percent in intervention factored in, the results are hardly reas- but they have ended up almost fully sites. This represents an appreciable suring for policy makers planning to subsidizing coverage. Countries such increase in enrollment but falls well achieve universal health coverage by as the Philippines and Vietnam have short of the sort of coverage rate that subsidizing voluntary health insurance not, and these are better tests of the would be considered universal. enrollment. voluntary enrollment model; unhap- In February 2012 a second experi- pily for the governments concerned, ment was therefore launched, focus- growth in enrollment among informal ing on the families in the “treatment” sector workers in these countries has municipalities that had declined the been sluggish. subsidy offer and had chosen to re- The question naturally arises as to main unenrolled. The aim was to see what additional steps governments what further steps beyond providing such as those in the Philippines and information and a subsidy might be Vietnam might take to raise volun- taken to raise enrollment rates. All IPP- tary enrollment rates and thus move eligible families that had opted not closer to universal health coverage. to enroll had their voucher extended, Surprisingly little is known about the were sent the enrollment kit a second relative cost-effectiveness of alterna- time, and received Short Message Joseph J. Capuno, Aleli D. Kraft, Stella Quimbo, tive measures. Service (SMS) reminders about the Carlos R. Tan Jr., and Adam Wagstaff. 2014. A new study by Capuno, Kraft, subsidy offer. Half these families, se- “Effects of Interventions to Raise Voluntary En- Quimbo, Tan, and Wagstaff presents lected at random, were given a further rollment in a Social Health Insurance Scheme: A evidence from two randomized experi- “treatment”: they were told that in Cluster Randomized Trial.” Policy Research Work- ments conducted in the Philippines, the end-of-study household interview ing Paper 6893, World Bank, Washington, DC. 6 World Bank ResearchDigest How Insecure Property Rights Affect Migration in China Land tenure insecurity reduced the ages 16–50 who worked as migrants suggesting less conflict, more land rate of migration from rural to urban during the period. rental, and more transparency in gov- areas in China during the 1990s and Establishing a causal relation- ernance in villages with a dominant ship between land tenure insecurity lineage group. Interestingly, leaders early 2000s and migration is not straightforward. of these villages are also more likely O Reductions in a farm household’s to face a challenger and lose a subse- ver the past 25 years China has land may simply reflect a choice by quent village election if they imple- experienced an economy-wide the household to farm less (or not at ment a land reallocation. The authors structural change as more than all) and to allocate labor to nonagri- posit that when residents are already 250 million of the country's farmers cultural pursuits. Outcomes of a land willing to transfer land through rental, shifted from work in small-scale reallocation initiated by village leaders land reallocations are viewed as un- agriculture to nonagricultural employ- may reflect unobserved bargaining be- necessary and village leaders are more ment. At the level of the household, or tween village leaders and farmers, and likely to be punished for forcing them family, the shift of labor from agricul- prior research has found that realloca- on the village. Establishing this rela- ture to industry and the commensu- tions substitute, though imperfectly, tionship, the authors use election tim- rate movement out of rural areas often for a market: households interested in ing and heterogeneity across villages proceed incrementally, with individual farming receive land from those shift- in size of the largest lineage group to family members migrating to urban ing into other activities. identify the risk of reallocation. or manufacturing areas while leaving Reallocating land is a costly exer- Demonstrating that reallocation other members behind. Through this cise for village leaders and can often risk is associated with a 17.5 percent gradual process family members be unpopular with some households, reduction in the probability that some- remain economically linked, and this either because they lose land or ex- one age 16–50 will migrate for work, arrangement often benefits the house- perience changes in their landhold- the authors provide evidence that hold by relaxing credit constraints ings or because they must spend time insecure property rights shaped the and providing an alternative source of and effort in the reallocation process. movement of labor out of agriculture. income. Further, any perceptions that a village The results suggest that strengthening The decision to migrate, however, is leader seeks advantage through the property rights over agricultural land shaped by institutional arrangements, process may erode support among vil- may be essential for removing biases both locally and in migrant destina- lage residents. The village leader thus in the movement of labor out of agri- tions, that affect the costs and benefits faces a potential political cost, reflect- culture during the process of structural of migration and off-farm employ- ed in the possibility of losing the next change and urbanization. As migra- ment. If poorly functioning institutions election. The authors suggest that two tion for work generally contributed to limit the functioning of land, labor, or additional institutional features, recent higher earnings and consumption for credit markets, they may raise or lower demographic change and the composi- China’s rural residents, one direction the expected benefits to individuals tion of lineage groups, also influence for future work would be to examine and households from moving out of the costs of reallocation and the pos- the effects of insecure property rights agriculture. sibility of losing the next election. on measures of income and well-being. In a new paper Giles and Mu ex- To estimate how the probability of As rural areas continue to depopulate amine how weakness in land tenure facing a land reallocation varies with with greater movement of labor to ur- security influenced the level and pat- election timing, and then the effect of ban areas, rural institutions, including terns of rural-to-urban migration in reallocation risk on individual migra- those governing land, can be expected China between 1995 and 2003. The tion decisions, the authors exploit to change in response to migration- paper develops a model suggesting data from retrospective household and induced social and demographic that the ability to improve claims to community surveys on elections, land changes. land, through residence in the village institutions, and individual employ- and continued work in agriculture, may ment histories that they conducted reduce the benefits of migration for in 2004 with the Research Center for farmers who want their households to the Rural Economy at China’s Ministry maintain a claim to land. The authors of Agriculture. These data were then find that risk of “land reallocation” in matched to household and village the following year reduces the annual panel data from four provinces (Anhui, John Giles and Ren Mu. 2014. “Village Political probability of migration out of a farm- Henan, Jiangsu, and Shanxi) spanning Economy, Land Tenure Insecurity and the Rural er’s home county by 2.8 percentage the period 1995–2003. to Urban Migration Decision.” Policy Research points, which accounts for 17.5 percent The authors first confirm evidence Working Paper 7080, World Bank, Washington, of the annual share of village residents from the political science literature DC. World Bank Research Digest 7 The Challenges of Public Service Delivery in the Pacific Islands The public sectors of Pacific island more than 10,000 people, for exam- resources available to them. countries are typically seen as being ple, is a quarter smaller than for the Returning to the paradox at the too big. But is this really the case, Caribbean. Moreover, both the number heart of the discussion, the paper sug- of inhabited islands gests that the size of given these states’ characteristics? and the geographic public sectors in the The size of public T spread of the Pacific Pacific island countries he Pacific island countries tend island countries—prox- sectors in the Pacific may well be appropri- to have large public sectors ied by the size of their relative to those in other small exclusive economic island countries may ate—or possibly even insufficient—when the states. Their public expenditure aver- zones—are more than well be appropriate— challenges of public ages 55 percent of GDP compared with 10 times as great. Their service delivery in these 34 percent for other small states, while population figures thus or possibly even states are taken into their public service wage bill averages understate the chal- insufficient—when account. Assessments 17 percent of GDP compared with 10 lenges they face with of specific functions percent for other small states. The economies of scale. the challenges reflect this when finding relatively large size of these public There has also been of public service that these public sec- sectors has led to the view that they inadequate recognition tors are too small: their are “too big,” an excess regarded as of how remoteness from delivery in these limited ability to take problematic both because it is unaf- major markets further states are taken into advantage of economies fordable (requiring unusually large inflates public sector of scale means that in aid flows to sustain) and because it costs, by pushing up account many areas of public “crowds out” private sector develop- the costs of inputs to administration there are ment (locking the economies into low public sector activity. The Pacific island simply not enough public servants or growth trajectories). countries are characterized by extreme resources available to carry out func- Yet the Pacific island countries are remoteness, much more so than the tions adequately. caught in a kind of paradox. At the Caribbean. same time as their public sectors are Using indicators of remoteness considered too big, assessments of from major markets and geographic any particular function typically recom- spread, the author constructs an index mend greater staffing to help remedy of remoteness and dispersion and inadequate capacity. How can their shows that these factors do indeed public sectors be both too big and too seem to play a role—alongside popu- small? lation size—in explaining the larger A new paper by Horscroft sheds size of public sectors in the Pacific light on this paradox. It examines evi- island countries relative to those in dence on economies of scale in public other small states. These factors also service provision and looks at how appear to explain some of the large particular characteristics of the Pacific variation in the size of public sectors island countries affect the costs of among the Pacific island countries. public service provision. These findings suggest that there is It is generally acknowledged that not a good case for treating the public small states require relatively large sectors of the Pacific island countries public sectors because they are un- as excessively large when the challeng- able to take advantage of economies es of public service delivery in such of scale in public administration. But small, remote, dispersed, and divided in the Pacific island countries too states are taken into account. The little attention has been paid to how findings also suggest a need to move internal dispersion and division exac- beyond the preoccupation with pub- erbate the problems these small states lic sector size in public sector reform already face with economies of scale. agendas in the Pacific island coun- The paper shows that their popula- tries. It would be more appropriate to Virginia Horscroft. 2014. “Public Sectors in the tions tend to be more thinly dispersed refocus those agendas on how effec- Pacific Islands: Are They ‘Too Big’ and Do They and territorially divided than those of tive and efficient their public service ‘Crowd Out’ the Private Sector?” Policy Research other small states. The share of their delivery is, given both their particular Working Paper 7102, World Bank, Washington, populations living in settlements of contexts for service delivery and the DC. 8 World Bank ResearchDigest (continued from page 1) Figure 1. Reported Interest in Migration, Search Effort, and Migration by Treatment Status Recent Policy Research Share of group (%) Working Papers 40 33.9% Control group 35 6960 The Domestic Segment of Global Supply Treatment groups: Chains in China under State Capitalism 30 21.7% *** All information Heiwai Tang, Fei Wang, and Zhi Wang All information + website 6973 Learning Dynamics and Support for Economic 25 15.7% *** Full assistance Reforms: Why Good News Can Be Bad 20 Sweder van Wijnbergen and Tim Willems 6977 The Dynamics of Centralized Procurement 15 Reform in a Decentralized State: Evidence 5.3% and Lessons from Indonesia 10 5.1% Audrey Sacks, Erman Rahman, Joel Turkewitz, Michael Buehler, and Imad Saleh 5 0.9% 1.1% 0.8% 0.8% 7012 School Autonomy and Accountability in 0 Thailand: A Systems Approach for Assessing Interested in Searched for work abroad, 2010–12 Migrated abroad, 2010–12 Policy Intent and Implementation Gustavo Arcia, Kevin MacDonald, and Harry working abroad Anthony Patrinos (full sample, at baseline) 7016 Barriers to the Implementation of Source: Based on data from baseline survey in early 2010 and endline survey in 2012. Environmental Policies at the Local Level in Note: Searching for work abroad includes asking family or friends, applying with a recruitment agency, applying online, or searching China another way. Sample includes all baseline respondents with completed endline surveys. Error bars indicate 95 percent confidence Genia Kostka intervals. Asterisks indicate that difference from control group is statistically significant at 10 percent (*), 5 percent (**), or 1 percent 7046 The Impact of Exogenous Shocks on (***) level. Households in the Pacific: A Micro- Simulation Analysis Caesar B. Cororaton and David S. Knight working abroad, and only 5 percent of Taken together, these results sug- 7054 Water Quality, Brawn, and Education: The Rural Drinking Water Program in China the control group searched for work gest that even in a context like the Lixin Colin Xu and Jing Zhang abroad over a period of more than 2 Philippines, where there is a lot of 7061 The Critical Mass Approach to Achieve a Deal on Green Goods and Services: What Is on the years (figure 1). infrastructure in place to support mi- Table? How Much to Expect? •  While many individuals lack informa- gration, unilateral efforts to facilitate Jaime de Melo and Mariana Vijil 7062 Performance of Renewable Energy Auctions: tion about wages and costs of work abroad, migration may have limited impacts. Experience in Brazil, China and India giving this information has little impact. The challenge is thus how to spur Gabriela Elizondo Azuela, Luiz Barroso, Ashish Khanna, Xiaodong Wang, and Yun Wu •  Alleviating frictions in matching demand for more migrants. Given the 7072 Diet Quality, Child Health, and Food Policies and barriers to getting passports leads to enormous development gains possible in Developing Countries Alok Bhargava more search effort, but no more migration. through international migration, in- 7091 Protected Areas and Deforestation: New Individuals who got all information vestigating policies that attempt to do Results from High Resolution Panel Data and the website assistance were three this appears to be an important area Brian Blankespoor, Susmita Dasgupta, and David Wheeler times as likely to search for work for research. 7119 Importing High Food Prices by Exporting: Rice abroad, but only 0.8 percent of them Prices in Lao PDR Dick Durevall and Roy van der Weide actually migrated. The full assistance Emily Beam, Dean Yang, and David McKenzie. package led to 21 percent searching for 2013. “Unilateral Facilitation Does Not work abroad and also had positive im- Raise International Labor Migration from the Working Papers can be downloaded at http://econ.worldbank.org pacts on receipt of job offers—but still Philippines.” Policy Research Working Paper To download the World Bank Research E-Newsletter, led to no overall increase in migration. 6689, World Bank, Washington, DC. go to http://econ.worldbank.org/research_newsletter The World Bank Research Digest is a quarterly publica- The Research Digest is financed by the Bank’s Editorial Committee: Indermit S. Gill (managing editor), tion disseminating findings of World Bank research. Research Committee and managed by DECRS, the Aslı Demirgüç-Kunt, and Shiva S. Makki. Editor: Alison The views and interpretations in the articles are those research support unit of the Development Economics Strong; production: Roula Yazigi. For information or of the authors and do not necessarily represent the Senior Vice Presidency (DEC). The Research Digest is free subscriptions, send email to research@worldbank. views of the World Bank, its Executive Directors, or the not copyrighted and may be reproduced with appropri- org or visit http://econ.worldbank.org/research_digest. countries they represent. ate source attribution. The World Bank 1818 H Street, NW Washington, DC 20433, USA Printed on Recycled Paper