FE43DF93-13CD-493E-BEF2-A77A9E7D6926 Copenhagen Consensus Center Logo
Copenhagen Consensus Center

Post-2015 Consensus: Science and Technology Assessment, Maskus

Assessment Paper

Summary of the most high-yielding targets from the paper

Science and Technology Targets Benefit for Every Dollar Spent
Expand open international circulation of skilled workers by 5-20% of current skilled migrants $15
Increase developing country R&D spending of GDP to 0.5%, and emerging countries to raise their ratios to 1.5% $3

Summary

The UN Open Working Group has considered a number of science and technology initiatives in its deliberations. It generally sees technology as a component of the objective “to strengthen and enhance means of implementation and global partnership for sustainable development”. Effective implementation of technical knowledge is fundamentally important to drive economic growth through structural change and productivity gains as well as contributing to broader public needs. However, most of the cases in which technology is listed in the working documents involve broad aspirations to promote and encourage technology transfer rather than specific targets.

We need objectives which both facilitate the broad targets of the SDGs while being reasonably amenable to numerical analysis. I have chosen two. The first is to increase the ratio of R&D spending to GDP to specific targets in developing and emerging economies, which was put forward by the working group but without any quantification. I posit that policies might be implemented to expand the ratio above the historical trend to either 0.5% or 0.75% in developing countries and either 1.5% or 2.0% in emerging countries. This would be achieved by expanding R&D tax credits to achieve 50% of the incremental target, with direct government expenditure accomplishing the remainder. Economic benefits are a spillover from additional R&D, but there are also additional costs.

The second objective is to establish 10-year visas permitting free mobility of skilled technical and managerial labor among participating ‘innovation zones’ in an effort to optimize technology diffusion, efficiency and learning. This would most likely require the (difficult) negotiation of plurilateral zones under the General Agreement on Trade in Services (GATS) of the WTO. I specifically consider the potential effects within two Western Hemisphere zones: North-South (Canada, the United States and the rest of the Americas) and South-South.

Benefits and Costs of National R&D Targets

Background and Literature

The benefits to private firms of investment in R&D do not in themselves justify government investment. The case instead rests on market failure: firms are not able to fully appropriate the returns to R&D investment and some of the knowledge generated spills over into higher productivity elsewhere, together with possible consumer gains. The social returns to R&D are higher than the private returns, with the case being clearest for basic research. Some of these gains are made across borders, making knowledge in one sense a global public good.

This presumably justifies the interest of the Open Working Group in raising R&D spending. But despite there being little doubt that R&D spending in OECD countries has contributed to growth, causal relationships are complex and difficult to sort out. Some studies have shown targeted tax credits can have a beneficial effect, while there is speculation that government funding of R&D in the USA has simply pushed up wages for scientists and engineers, who are in limited supply.

Current Situation

The most recent GNI-weighted ratio of R&D to GDP among developing countries (excluding low-income countries, where spending is very low) is 0.19%. This grew by about 3.7% annually in the first decade of this century. If it continues to grow at the same rate, the ratio would reach 0.3% by 2030. Among emerging economies, excluding China, the weighted ratio is 0.83%. This has grown at about 2.8% a year, which would imply a figure of 1.18% by 2030 if the trend continues.

China has a current R&D ratio of 1.69%, which has been rising at a remarkable 9.5% annually. At that rate, the ratio would reach 4.1% by 2030. Developed economies have ratios from 1.25 to 4.54%, with a weighted average of 2.61%. This has grown at only 1.3% annually, a reflection of the diminishing returns as the technological frontier is approached.

I assume the rising trend will continue at the same rate and the cost-benefit analysis must therefore focus on policies which raise the growth rate further. I define both modest and ambitious objectives: for developing economies, an increase from the trend-defined 0.3% to 0.5% in the conservative case and to 0.75% in the more ambitious one. This implies annual growth rates of 11.0 and 19.7%. For emerging economies, the targets would be ratios of 1.5% and 2.0% of GDP, and corresponding growth rates of 5.4 and 9.4%. Given existing and projected ratios, there seems little reason to include China and the developed nations in any program to increase R&D spending still further.

Models and Assumptions

I focus on a combination of two policies, which are assumed to operate independently. Based on existing data for developing and emerging countries, I assume that future increases will be split evenly between public and private sources.

I use an adapted partial-equilibrium model of R&D costs and benefits, developed for Canadian policymaking. I assume the policies begin in 2015 and have effects from 2016 to 2030 and also that real GDP grows at a constant rate (5% for developing economies and 3.5% for emerging ones). The policies used are direct government grants to R&D and an R&D tax credit for the private sector.

There are three factors on the cost side of the model: the efficiency loss imposed on the economy by subsidising the use of capital and skilled labor for R&D, the marginal efficiency burden of raising taxes to pay for the fiscal cost of increased public expenditure and the costs of government administration and enterprise compliance. There are two domestic benefits of these subsidies. One is the spillover productivity gains to the economy and the second is the possibility that spillovers will generate additional domestic innovation via reverse engineering, demonstration effects and learning by doing in laboratories.

In addition to the external productivity impacts on the supply side in investing countries, greater R&D expenditure has the important benefit of supporting external consumer gains in the home economy. These benefits can also cross borders, particularly from publicly funded basic research in areas such as agriculture and healthcare.

As an illustration of the inputs to the model, for the group of DCs, the targeted additional R&D spending would rise from $1.6 billion in 2016 to $30 billion in 2030 under the ambitious growth scenario. For ECs, the rise would be from $11.8 billion to $257.5 billion over the same period. In Canada, the direct spillover from incremental R&D investment in enterprises was estimated as 0.56 (56 cents per dollar invested) and 0.76 for direct public investment.

Results

For the ‘modest’ target of 0.5% share of GDP for R&D in developing countries by 2030, program benefits amount to $18.7 billion for costs of $17.2 billion, giving a benefit-cost ratio of 1.09 (at a 3% discount rate). Assuming a continuation of domestic gains for ten years beyond the 15 period of investment raises the BCR slightly to 1.13. The most favourable situation is where the induced innovation benefit rate is increased to 0.04, giving benefit-cost ratios of 1.2 over 15 years or 1.4 over 25 years. Adopting the more ambitious target of 0.75% of GDP increases both benefits and costs, but the BCRs are still in the same range. For emerging economies, although benefits and costs are on a much larger scale, there is essentially no change to the rather low benefit-cost ratio.

If the primary goal of programs designed to boost the relative share of R&D is to expand domestic productivity, these seem a relatively poor investments. Developing and emerging economies may be better advised to invest in human capital and governance to establish solid frameworks within which knowledge spillovers from access to international technologies are optimized.

However, the situation is different if there are large usage spillovers. Over 40 years, the additional domestic and international consumer gains from an 18% spillover rate would raise discounted benefits to between $32 billion and $46 billion for developing countries. Since program costs would essentially be unchanged, the benefit-cost ratio rises to a range of 2.3-2.7. With an assumed 30% spillover rate, BCRs are in the range 3.2-3.9. Ratios are similar for the emerging economies.

The attractiveness of such policies depends therefore on the views of program goals and likely outcomes. If the emphasis is on expanding the access of poor countries to international knowledge, this approach does not seem an attractive option, although the actual benefits may be somewhat understated. However, if significant external consumer gains could be produced and shared across borders, then net benefits could be more interesting.

Benefits and Costs of Expanded International Skilled Labor Mobility

Many international economists argue that the actions that would most increase global efficiency and welfare, while raising incomes in the developing world, are significant relaxations of barriers to cross-border migration, including temporary movements of labor. The simulated welfare gains reported in the literature  are staggering, ranging from 67% to 147% of global GDP. In comparison, removing all remaining barriers to trade and capital flow would each only amount to about 1% of world GDP.

A complete elimination of migration barriers would equalize wages as a point of equilibrium is reached between labor supply and demand in both rich and poor countries. Wages in countries with net emigration would rise as the labor supply diminished, while they would fall in countries with net immigration. Migrants would see enjoy a large real income gain, while workers who did not migrate would also see a rise in wages. Owners of capital and land in the source country would see some decline in income as productivity fell. In the host country, workers lose some real income, but owners of capital and land gain productivity. The net global welfare rise goes largely to the workers who move abroad. However, these workers become as productive as those in their destination country so that global productivity and welfare also increases.

This is not the whole picture, of course. Whether or not the source country becomes worse off depends on whether emigrants decide to remit some of their income. The analysis may also be complicated by the fact that incoming workers are actually inherently less productive than those in the host country. Another possibly important factor is that of welfare externalities, which form the basis of concerns about the ‘brain drain’. Public investment in education may be lost when skilled workers move abroad and their departure also reduces the remaining tax base. The presence of technically skilled workers in the host country, on the other hand, may generate greater efficiency in other workers and firms through innovation, learning and attraction of FDI.

A final observation from the literature is that the very large theoretical global welfare gains are for a situation where there is complete labor mobility. This is impossible, both in political and economic terms. Nevertheless, more restricted movement can still have significant benefits; a 7% net emigration rate might generate an extra 10% growth in global output and even 2% could achieve a 2.4% rise in world GDP.

Current Situation

Figures on the numbers of people with tertiary education from a range of source countries then residing in OECD countries are available for census years 1990 and 2000. This enabled me to estimate the numbers of foreign-born immigrants and permanent residents with higher education in 2000 in Argentina, Brazil, Chile, Mexico, Canada and the United States plus aggregates for Central America, the Caribbean and the rest of South America.

This captures migration from high- to low-income countries as well as in the other direction. I also made estimates of flows not recorded in the database, such as from the Caribbean to Mexico or Central America. Since the figures are for all people with higher education, I then estimated the share of managerial and technical workers (MTW) in each source. Using these figures together with data from the UN on bilateral stocks of foreign-born legal migrants, I was able to update these estimates to 2010.

There is some movement between all countries, but it is no surprise that Canada and, especially, the United States are large net destination countries. The USA had, by these estimates, 1.9 million migrant MTWs in 2010, with over 700,000 of these coming from Mexico. Interestingly, America also supplies by far the largest number of MTWs to Mexico. There are proximity effects, with Chile, for example as the largest source for migrants to Argentina. Despite its large economy, Brazil is the smallest gross recipient of migrants.

Innovation Zones: The Idea

The straightforward policy proposal is to create ‘innovation zones’ (in this case the Americas) within which eligible skilled workers could circulate freely by obtaining a work visa from an agreed increased supply. I consider two types of visa relaxation for North-South and South-South innovation zones.

Model and Assumptions

In this model, I estimate welfare benefits and costs for visas issued from 2016 to 2030, with economic impacts extending to 2040. I assume visas are increased by 5% of the initial MTW inward stock, with demand across source countries in proportion to their existing share. Estimates are based on initial bilateral mobility. Because the proposed free movement could result in more workers spending time in Canada or the USA, where wages are highest, the BCRs are likely to be higher than estimated. However, costs in poor countries may be higher because fewer workers would choose to go there.

The largest benefit is the real income gain to those who become mobile. Although there is little evidence to suggest that immigrant workers are intrinsically less productive, I use the conservative assumptions that the productivity factor is either 0.2 (low value) or 0.5 (high value). Although migrants from low-wage countries will increase their earnings, I assume that movements from high- to low-wage economies are not accompanied by any wage loss. Because wages in developing countries are likely to grow faster, I permit a 2% catch-up between low-wage sources and high-wage destinations in a year. Although remittances have no effect on the overall within-zone welfare benefit, they alter the distribution significantly. I use a low estimate of 0.2 and a high of 0.32 (based on literature reports) for the remittance rate.

In considering the brain drain controversy, I assume that wages are taxed in the destination country but that remittances are only taxed at their destination. Although there will be some tax losses, the free movement of skilled labor may actually increase tax revenues in poorer countries as higher-paid migrants move in. Welfare impacts then depend on any possible reduction in taxation of income, which I assume has to be replaced by increasing revenues from other sources.

The human capital impacts of the brain drain are also important. However, no evidence has been found of significant human capital externalities; for example African countries experiencing the greatest outflows of medical professional have systematically better health characteristics. Indeed, there may even be an endogenous ‘brain gain’ as more young people are induced to invest in schooling by the prospect of a move abroad. Although I am sceptical about the conclusions of these arguments, I have added a cost factor for the North-South innovation zone to capture additional efficiency losses.

Technological spillover and dynamic effects are potentially more relevant, as for the case of R&D spending, but the big question is the extent of this. I assume that each dollar of high-technology imports, FDI and licensing raises local TFP by $0.03 and that the increased flows of managerial and technical workers allow half of this to be captured. For each $1 billion earned in wages by MTW migrants, there is then a real GDP gain of $15 million. This is the benchmark for North-North movements, which I scale for other directions.

Note that greater openness to movement of skilled labor could give a significant boost to growth, in a similar way to that seen for trade liberalization. Even a small increase would dramatically expand benefits and benefit-cost ratios.

Results

A 5% increase in bilateral flows within a Western Hemisphere innovation zone would result in a circulation of about 136,000 workers, compared to a current cap in the US of 65,000 H-1B visas. For a conservative set of parameter estimates, there would be large gains in income to the MTW movers ($16.3 billion), with about $3.2 billion remitted to source countries. In terms of tax-related fiscal externalities, source countries lose about $2.2 billion in welfare while destinations gain in aggregate $2.8 billion, with most of the developing countries suffering net losses on this account. Total program benefits far exceed costs for all countries except Canada and overall benefit-cost ratios in developing countries range from 5.8 in Chile to over 47 in Central America. However, most of the gains are actually to workers rather than to source countries directly. The overall BCR for the program is 9.3, placing it in the ‘phenomenal’ range. Taking out the net benefits to workers reduces it to a more modest 1.8.

The set of medium parameters – high-end technology spillovers with low-end productivity gains and remittances – raises the BCRs a bit and developing countries enjoy somewhat larger net benefits. The full program BCR is greatly increased (to 21.3) by using the high end parameters, with income gains of movers increasing to $41 billion. Allowing for some of the negative consequences of the brain drain reduces the ratio to 18.3. The scale of economic effects for a South-South innovation zone is far smaller, even with high end parameters, with the benefit-cost ratio falling to 2.9.

A much larger scale scenario is to permit 20% increases in visas, phased in linearly over five years and applied across the North-South zone. In-zone mobility would amount to over half a million managerial and technical workers, with nearly 390,000 of these going to the United States. Circulating professionals would gain an NPV of over $158 billion over a 25 year period, discounted at 3%. Considerably larger welfare impacts from tax losses would be more than offset by destination country gains. Potential gains in GDP from technology spillovers would amount to $6.9 billion, with about three quarters of this going to Mexico, Canada and the USA. Costs and benefits rise nearly in proportion, still giving an overall ratio of around 21.

Further Remarks

In my view, the net gains are likely understated for at least two reasons. First, the literature points to a complementarity between the mobility of skilled workers and volumes of technology flow. Second, there may be a growth dividend in line with experience on trade liberalization. Speculation about a global (rather than regional) innovation zone, based on that experience, reasonably could anticipate that the costs and benefits would more than double, with ratios little affected.

Concluding Remarks

Raising the share of R&D in overall GDP has been shown to have benefit-cost ratios rarely exceeding 1.4, giving this a low ranking compared to other SDGs. However, the second option of forming innovation zones is much more attractive, with a benefit-cost ratio of about 21.

It could be argued that the relatively small contribution of TFP spillovers makes innovation zones an inefficient means of gaining more technology. However, the increases are notable and form permanent increases in learning and network connections, and my analysis has given a conservative estimate of the gains. And, of course, the additional openness to the mobility of skilled labour is a direct means of access to technology. I suspect also that endogenous affects could also be large and are certainly worthy of further study.