The Elusive Quest for Growth, William Easterly, 2001, Cambridge: MIT Press
The Harrod-Domar Model was inspired primarily by pre-war experience in the US and Soviet Union. In both cases, a strong link was perceived between physical investment and output: in the Soviet case, the ability of the government to mobilise high levels of gross saving and invest this in physical machinery was seen as a key reason for the extremely rapid rate of industrialisation; and during the Great Depression the obvious superfluity of workers seemed to imply that it was capital that determined the number of them in work (even if capital sat idle during the Great Depression). Domar himself developed his model in order to explain certain esoteric features of the business cycle in developed countries, and later firmly disavowed the use of his model in explaining the determinants of the long-run rate of growth of poorer nations.
The basis of the model is extremely simple: it is the assumption that there is a more or less fixed ration between one period's total investment and the following period's economic growth. The approximate factor assumed often seems to be four: so if investment is 10 per cent of GDP in this period (this period perhaps being four years) then economic growth in the following four years will be 2.5 per cent per annum.
The next observation is that saving is low in developing countries, because few individuals have the money to save. And so The Solution: provide aid to fill the 'financing gap' between savings and investment. The World Bank wrote a computer programme to do this and voila: a computer spat out the aid requirement of each country based on their growth requirements. This model fell out of favour in academia in the late 1970s and has been viewed as thoroughly discredited ever since, although almost no empirical testing has been done on it. Easterly runs a few regressions and finds:
A modified and weaker form of the theory then claimed that investment is a necessary but insufficient condition for growth, so Easterly adds a third test, which finds
The extent to which the data fails to support the Harrod-Domar Model is quite surprising.
The main reason Easterly offers is that the model ignores incentives. The World Bank's means for determining aid requirements rewarded a lack of investment — it was countries that had the highest financing gap that got the most aid, so saving was discouraged and aid was used for consumption. He further claims that there is a weak and complex association between investment and growth. Investment in machinery needs to be motivated by a concern for investing in the future if it is to be accompanied by all of the other things that are required to make growth happen (investment in people and institutional design and adaptation of technology). Despite the Harrod-Domar Model's failure, it persists in influence to this day, particularly in the IFIs. It has fallen out of favour with academia, and yet casual statements of the need to increase investment in order to sustain or improve growth slip through many forums of discourse without challenge.
The term capital fundamentalism refers to the belief that the accumulation of physical capital is the principal determinant of long-run growth. It was first challenged by the Solow model presented in two papers in 1956 and 1957 which argue that diminishing factor returns imply that growth from capital accumulation can only exist in the short run, during a transition period. Instead, the unique driver of long-run growth is technological change.
This view was widely and rapidly accepted in economics. However, its typical application to development economics was spurious. It was commonly held that there were insignificant barriers to the movement of technology across borders, and therefore that the difference in wealth between countries could only be due to a difference in levels of capital accumulation — that poor countries were beneath their long-run growth trajectory because of their lack of capital accumulation. One corollary of this is that the wealth of countries naturally converges: if all countries have the same technological base then those with the scarcest capital will offer the best returns to investors, so investment will pour in and tend to equalise the level of capital accumulation.
This view of reality fails all empirical tests. It fails to explain how rich countries became rich in the first place — which can only have happened through a process of divergence. It fails to explain why rates of return on capital are so low in the developing world. It fails to explain why labour productivity is so low — the absence of capital is not quantitatively sufficient. It fails to explain the further divergence observed throughout the world since 1960. It fails to explain the wide variety of growth outcomes stemming from similar rates of capital accumulation in different countries.
The central message of Solow — that long-run growth is first and foremost a product of improvement in technology — was lost on development economics for a long time. Technology is not as mobile as it was cheaply assumed to be. Capital fundamentalism persists to the present in reports of the IFIs and UN, even though the idea has by now been thoroughly discredited.
Education has been touted as a panacea for decades, often by highly overeducated individuals. However, a strong body of evidence suggests that educational expansion is associated with growth. The few studies that exist have serious weaknesses, particularly that of causation: one would expect an increasingly affluent society to increase its investment in education.
Easterly suggests three main explanations as to why education has failed to deliver the economic growth that so many expected. The first is the use to which educated people are putting their skills. Many countries in which government intervention in the economy is extensive create a variety of opportunities for well-educated people to profit by lobbying government or working the system. For example, foreign exchange controls tend to create an opportunity for intelligent black market currency trading, at the expense of government and importing or exporting firms. A second reason is the way in which education has been provided, generally by the state and often compulsorily. The quality of education is often very bad. Teachers are underpaid and unmotivated, and students are not provided with basic materials like books and pens that they need to succeed. In some context teaching posts are a form of political patronage and teachers are overprovided whilst nothing is spent on basic teaching materials. The final reason is other trends within the economy. Unless there is investment in machinery and technology that is generating a demand for those skills then the creation of these skills will have no economic impact. It ought to be true that the generation of a high-skill workforce creates incentives to invest in machinery and technology, but it is equally possible that government policy has more than offset this incentive.
There has been an enormous international effort aimed at providing free contraceptives to the entire developing world. It is based, in Easterly's view, on the assumption of a negative relationship between population and per capita growth — that, in a sense, GDP growth is not affected by changes in population and there is a fairly direct relationship between per capita income and population size. This presentation often seems a little dishonest. For one thing, even in his theoretical explanation of the economics of the situation, Easterly does not mention fixed factors. He uses the rapid increase in agricultural productivity in the post-war period to deny that land should be regarded as a limited resource, not mentioning the dependence on this expansion on another fixed resource, oil. He does not mention water as a fixed resource. He does not mention the likelihood that these constraints, particularly oil and water, only become significant when they bind, as they seem likely to do in the first half of the twenty-first century.
He argues that the fact of large numbers of unwanted births is a myth, that 90 per cent of the variation in fertility across countries is accounted for by variation in desired fertility. 10 per cent seems like an awful lot to me, but apparently not to him. He argues as an economist that the decision of how many children to have is far too important to be significantly influenced by the subsidisation of condoms, which would be supplied cheaply by the market if they were desired. He argues that it is impossible for people to be unable to afford contraception if contraception is cheaper than having children. He notes that various authors have predicted famine and disease to result from population growth in the last few decades, primarily Paul Ehrlich,1) and that these disasters have not transpired. He also summarises a literature that finds no association between population growth and per capita GDP growth. There is no empirical evidence that population growth is harmful. Furthermore, by far the most robust finding is that increase in income reduces fertility, that the most foolproof method to reduce population growth is to increase economic growth (again this ignores the view that both population and economic growth push us towards binding resource constraints and in this respect might be equally problematic). He briefly discusses the externalities of having children as a justification for subsidising contraception, concluding that there are both positive and negative externalities and it is difficult to conclude that one set easily outweighs the other.
HIV is not mentioned once, not even from a purely economic standpoint.
Partly in response to the debt crisis in 1982, the IFIs made a drastic policy change from providing aid primarily for individual projects to offering longer term loans to support government budgets more broadly. These loans were made conditional on changes in policy, in the hope that this policy reform would prevent a recurrence of the debt crisis and promote growth. There were some successes (eg Ghana, Thailand, Korea, Argentina!) in which loans coupled with policy change lead to improved growth in comparison to the previous period. But in many more cases loans were repeatedly disbursed without a noticeable shift in policy. Governments developed an enormously diverse means of giving the appearance of reform without improving policy, particularly by using a variety of accounting techniques to steal from future assets to avoid current deficits. In other instances, governments would enter a recurring pattern of negotiations with donors in which reform would be attempted, aid would be accepted and policy would slide back, so that the government could reapply for the same loans all over again. Empirically, IMF loans did not lead to increased growth or improvements in policy.
Easterly explains this failure in terms of a range of perverse incentives on the part both of donors and recipients:
One study claims that 1 per cent in GDP of aid in a good policy environment leads to 0.6 per cent growth in GDP, implying that this failure of structural adjustment loans to achieve an improvement in policy has cost the developing world a great deal of growth.
Easterly's recommendations are simple. Firstly, the amount of aid offered should be conditional primarily on past performance, ie proven track record: on key indicators of policy (inflation, black market exchange rate premium) and growth. Secondly, aid should increase as countries become richer and as the numbers in poverty decline. This is absolutely necessary to avoid providing perverse disincentives to growth and poverty reduction.
Clearly, the obvious objection to Easterly's proposals is that he argues that most aid should be given to those who need it least, and no aid should be given to those in desperate need. His assumption appears to be that the really crucial determinant of growth is policy, not aid, and that growth and poverty reduction can be achieved with good policy and no (or very little) aid. He believes that the empirical evidence bears him out: that aid is only beneficial in a good policy environment, so it is better to have good policy and little aid than lots of aid being wasted by bad policy.
Debt forgiveness has been a routine feature of the donor-recipient relationship since at least 1979. Significant amounts of debt have been forgiven throughout this period. However, empirically, debt relief is usually accompanied by a greater level of new borrowing, and other means of mortgaging the future such as artificially high real exchange rates (which favour consumption of cheap imports over the long-run development of an export sector) and selling off state assets (even though Easterly acknowledges, bizarrely, that privatisation is typically an IFI condition of debt relief). This proves that debt relief is given to profligate governments who have a preference for high present consumption and a high degree of indebtedness who have no intention of changing policy. There is similar evidence that high borrowing is a consequence of bad policy rather than bad luck. Debt forgiveness is therefore a means of subsidising bad governance, and an incentive to borrow and remain highly indebted. Debt forgiveness does not help the poor, but merely enables bad governments to stay in power continuing bad policies. Paradoxically, the withdrawal of such relief would be preferable from the perspective of the poor, as it would force changes in governance. Easterly illustrates with a case study about Cote D'Ivoire, which used its high indebtedness to gain far greater support than similarly poor countries with better policy, such as India, where aid would have been better used.
Debt forgiveness should therefore only be undertaken when two strict conditions can be met:
In particular, the notion of the 'financing gap' must be 'abolished, now and for all time,' and the notion that donors must lend to fill a shortfall in a government's ability to raise revenue itself has to end. The only way to incentivise governments to alter policy so as to generate sufficient revenues themselves is to leave them to deal with the consequences when they don't.
Easterly acknowledges, albeit briefly, that bad lending was as much to blame as bad borrowing. However, he avoids the debate as to whether there is a legal or moral case for demanding repayments of debt lent to a corrupt government by a population that never gave any kind of consent to the borrowing. In the case that a donor has chosen to lend to a government that cannot claim to represent its people, that uses the money in a way that does not benefit its people, then I feel that there is no legal or moral basis on which international donors can demand repayment by a subsequent administration. That's a basic principle of capitalism: the borrower repays, and if you can't extract repayment from the borrower then tough luck, you shouldn't have leant in the first place. If loans don't lead to an expansion of the productive base that enable them to be easily repaid, then chances are they were badly selected loans which the lender has no right to expect repaid. The implications of this argument are that debt forgiveness will in many cases be a right of poor countries, but there can be no expectation of new lending where governance remains bad. These governments will have a preference for high indebtedness, but within a legal framework in which creditors cannot expect repayment from a more representative successor, they will have no opportunity to borrow since nobody will be willing to lend.
Traditionally, economics assumes diminishing returns by default, and has found the possibility of increasing returns difficult to work with. Therefore, it is assumed that a high-skill individual in a low-skill society will be of greater value than if he were in a high-skill society, because of the scarcity of skill as a resource.
However, the contrary view is that high-skill workers require other high-skill workers to be productive, that economic production requires collaborative organisation and collaborative organisations often require a lot of people with similar skills or similar levels of skill. Therefore a high-skill worker can be more productive in a high-skill economy because it is easy to find people to collaborate with, whether that be close collaboration (setting up a business together) or distant collaboration (providing basic arms-length services such as efficiently functioning telecommunications systems). This implies that there are increasing returns to skills for an economy — the more skills it already possesses, the greater marginal return to an increase in skill level. Easterly calls this 'matching,' in the sense of high-skill workers benefiting from 'matching up' with other high-skill workers.
A similar situation exists with knowledge, technology and the embodiment of technology and knowledge in machinery. The traditional diminishing-returns story argues that machinery will yield the greatest returns where it is most scarce, in low-capital environments. But it is also possible that in some instances, incremental investments in knowledge will be most valuable in an environment in which a lot of knowledge already exists, because knowledge can be complementary: two pieces of knowledge, when combined, can yield a greater return than the sum of using them separately. Many new ideas in developed countries can only be used because of the already accumulated knowledge and technology which enables us to put them into effect. Easterly refers to this process of knowledge complementing other knowledge as knowledge 'leaks'.
In both cases the traditional diminishing returns still exist and are important; the mechanisms by which returns are increasing are separate and in competition with diminishing returns. It is unclear which will dominate in a new situation.
Where increasing returns dominate, there is a tendency for increasing returns to create 'traps,' that is, virtuous and vicious cycles. In a low-skill economy, the returns to investment in education will be low because of lack of other high-skill workers with whom to cooperate to fully exploit one's skills. The incentive to seek education may therefore be too low to be worth the costs involved, and it will be rational to remain uneducated and poor. It might only be worth investing in education if it is possible to migrate to a high-skill economy to find these matches — so brain drain tends to exacerbate the problem.
Easterly extends this idea of traps to explain variation in income by geography and race purely on the basis of a difference of initial conditions, because under increasing returns variation tends to be self-reinforcing. He proposes a race model in which whites begin as high-skill and blacks low-skill. If there is any segregation in economic enterprise between the races then it may be rational for whites to seek education and blacks to not. Furthermore, if there are costs associated with uncovering the true educational level of a particular worker, it may be rational for a white employer to only select whites rather than pay to discover the true educational level, on the basis that the probability that a white his high-skill is high and for a black, low. If white employers are known to use this policy, then it will be rational for whites to pursue education and for blacks not to: even without a legal or geographical divide, the perpetuation of a racial income disparity can be rationally self-reinforcing. This is a broad, powerful and intuitive 'it's-not-their-fault' defence of low income groups, races, countries and geographical areas.
Finally, expectations matter. In deciding whether or not to invest in education, it is not the current state of national skill levels that are important, but expectations of future skill levels when the investment in education comes to maturity. There may therefore be a coordination problem in which many people are willing to invest privately, so long as the others do likewise, though this is generally impossible to negotiate or enforce. But credible promises by some competent agency may be very important.
The immediate policy implications of this analysis are reasonably straightforward.
This is a plausible story of the government-business collaboration that helped jump-start the East Asian growth miracle. —p169
Technological progress is not smooth and homogenous, and it is necessary to look a little closer at the modalities of technological improvement to determine whether technology tends to narrow or to widen the gap between rich (technologically advanced) and poor (technologically backward) countries. In particular, it is important to factor in the role of incentives: to examine when incentives exist to rapidly adopt new technology and when there are stronger incentives to resist technological progress.
For the reason of non-appropriability and obsolescence, the rate of technological innovation will tend to be too slow in a market economy. These disincentives to innovation can be so strong that there is no innovation and thus no growth in a free-market economy. The way out would be to create strong incentives for innovation by subsidising private research and development, subsidising adoption of best-practices foreign technology, encouraging foreign direct investment from high-tech places, having the government itself do some research and development, and having strong intellectual property rights that allow inventors to keep the profits from their invention.2) —p178-9
The first case is where new technology is a net substitute for old. In this case, there is a natural incentive for vested interests to resist rapid adoption of new technology. Such vested interests may be producers (companies and unions), but they may also be consumers, if the new technology requires them to learn new skills (for example, consumer resistance to newer open-source software which required more effort than continuing to use existing closed-source systems). Substitute technologies tend to result in lurching technological progress, that often moves from one place to another. An outdated industry can hold out against a new technology for a certain period; then the competition becomes overwhelming and it collapses. Production moves from a country that has mastered the old technology and is resistant to the new to a fresh site where there is no resistance to the new methods. Example: steel production moving from Britain to the US to Japan to Korea, each move based on a new industrial technology. Where high capital investment encourages gradual, incremental changes in production methods, there is an inherent technological reason to resist the adoption of a drastically new production method until the old equipment can no longer be maintained.
Substitute technologies favour backward countries and encourage convergence between the rich and poor. It is easier for a country with no existing industry to adopt technology on the frontier than one with a resistant domestic industry. The ability of legacy technologies to resist the new may also depend on the relative power of the old and skilled in society or in an industry relative to the young and unskilled. Growth fuelled by rapid technological adoption is often associated with a young population (SE Asia), or with a society emerging from a war which crippled the power of established leaders (Europe and Japan after World War II). Imitation adds to this process, where technology can be easily copied by poor countries without having to invest in research and development themselves.
However, all of this is dependent on the details of the technology involved. New technologies can also be complementary to old, making existing technologies more efficient. Each new piece of software increases the usefulness and efficiency of computers and networks. Moreover, technology is not only complementary with itself but also with skills — many new technologies are much more valuable in the hand of already skilled people, even people with completely general skills rather than skilled in the particular form of new technology. Telephones are more complementary to professionals than to farmers. On top of this, innovation is more likely where technology already exists in abundance, amongst highly skilled people. Where technologies are highly complementary, there will be a strong cluster effect, drawing skilled people and technology together into stable production hubs — they will exacerbate existing inequalities.
Finally, separate to concepts of complementarity and substitution is the importance of luck — or 'path dependence'. Some technologies (although, crucially, this is probably not apparent to begin with) naturally lend themselves to further improvement and refinement whilst others don't. Examples: in Europe, the use of wheelbarrows gently evolved through carts to railways whereas in the Middle East, camels were more useful to begin with but were much more difficult to refine.
To some extent which effect dominates — the convergent or divergent — depends on whether poor countries can identify and invest in those technologies which naturally offer an advantage to those without incumbent resistance.
There is a chronic tendency for economists to ignore the role of luck in the differential success of economies, even though it is capable of explaining a significant part of growth outcomes.
It is worth mentioning the concept of 'mean reversion', by which the most predictable outcome in a sample dominated by randomness is that the highest performers will do worse in the next period, and the worst performers will do better.
The principle of mean reversion is universal. All you need to get strong mean reversion is at least some role for luck and selection of the best outcome of the previous period. Mean reversion explains why the Rookie of the Year in the American League has a worse second year (the so-called sophomore jinx — the Rookie of the Year moves back toward the average after an exceptional first year), why the NFL Super Bowl winner seems to fall apart the next year (the team doesn't really fall apart; it just falls back toward the mean), why second novels are disappointing (we pay attention to the second novel only when the first was exceptional), why movie sequels are usually not as good as the original (a sequel is made only after an extremely successful movie, an extreme success is unlikely to recur), and why a stock market prognosticator falls out of favour right after a streak of accurate predictions (she had a lucky streak that got our attention and then reverted to average). —p205
In Search of Excellence provides a good example — a survey of thirty-six exceptional companies, around two thirds of which performed below average over the course of the next period.
I don't really believe that growth is completely random. I hope that evidence elsewhere in this book will convince you that government policies and other factors have a strong association with growth and prosperity in the long run… Keeping in mind the role of luck in economic development will… allow us to be more charitable toward countries where growth has taken a dive. Bad government policies are usually partly to blame, but so is bad luck. —p214
There are well established associations between stylised 'bad government policies' and low growth. Unfortunately, causality is far more difficult to demonstrate. In some cases, some (much weaker and more conditional) evidence exists that there is some proportion of causality, although it's far from clear that a sizeable portion of the association is not a consequence of governments faced with poor growth turning to bad policy either out of necessity (in the case of, say, inflation and budget deficits) or desperation. That said, here they are:
create the anticipation of future tax hikes to reduce the deficit and service the public debt. They raise the possibility of inflation that will tax money holdings. They lead to general macroeconomic instability, which makes it hard to tell which projects are good and which firms should get loans. —p226
This seems to be one of the weaker arguments, in particular because a high budget deficit is such a natural consequence of negative external shocks (eg terms of trade shocks) that Easterly has already identified as a growth killer. He spends some time on a single case study — Mexico — which seems to start from a sound economic position, adds budget deficits, and ends in collapse, but whether that story generalises is a little more unclear.
He goes on to temper his previous statements, explaining that it is difficult to define exactly which components of openness are important, that different studies have picked out different components with varying degrees of success, and then — unlike in other sections — notes that critics such as Francisco Rodriguez and Dani Rodrik have largely undermined these results by changing the specification.3) “What does hold up well,” he concludes, “is that the whole set of policy distortions of free trade is negatively related to growth” — in other words, some or many restraints on free trade may be a good thing, but a very closed economy is going too far. It is a little unclear whether economies that are closed due to geographic conditions are included in this finger-wagging.
Finally, the “missing policy” — income taxation — “should” be here (according to sound economic logic) but isn't, since no evidence has been found that it negatively affects growth.
Corruption is insufficiently addressed by the literature (eg Ray5) does not mention it at all). Corruption harms growth, acting as a tax on production. Corruption exists in every economy, without exception, but varies widely in scale.
There are two main varieties of corruption:
Determinants of corruption are mostly fairly obvious. Ethnic diversity is associated with higher levels of corruption, and where ethnic diversity exists, more foreign aid is associated with more corruption (though not in ethnically homogenous environments). High black market premia are associated with high corruption. Almost tautologically, high corruption is associated with low institutional capacity (which Easterly divides into four):
These are almost depressingly vacuous:
The bulk of this chapter is an astonishing glut of anecdotal evidence of interethnic conflict, for some reason. His basic argument is that a divided population will face an open-resource problem in which competing groups vie for resources and public spending, precluding the consensus required for rational investment in growth-promoting policy. He names the two most common and important divisions as:
Racial segregation tends to lead to competition over which community gets the benefits of public spending — and has a tendency to invest more where the beneficiaries of that spending are of a homogenous group and less when the beneficiaries are varied. Government tends to change rapidly and suddenly from one ethnic group to another, causing disruption in development programmes. In the worst case, ethnic division causes war and genocide, which, lest we forget, are awfully bad for growth. He also explains 'policy attrition', in which multiple groups refuse to make a sacrifice in order to implement good policy, each hoping that the other will blink first and make the sacrifice themselves. He seems to associate this with inflation in particular. He also notes that foreign assistance, when no care is taken to ensure different ethnic groups benefit equally, can easily exacerbate the problem.
Class conflict causes a tension between two possible government aims: redistribution and growth, which are, Easterly strongly asserts, in general mutually exclusive. When inequality is above a certain threshold, the disenfranchised classes will see a greater benefit from putting their efforts into redistribution than they would see from growth. Predictably, the most problematically polarised societies are split by both racial and class lines, and it is fairly common for ruling and business classes to be of a different ethnic group to the remainder of the population, creating very poor incentives for politically powerful groups to enact policies which enhance the position of ordinary peasants and workers, since this would threaten their privileged position in the longer term.
Easterly's policy prescriptions, as for corruption, are a little vague and woolly:
A summary of policy prescriptions, each focusing on incentives: