Lecture 1 Economic Development
What is Economic Development?
While this may seem like the basic question we need to answer before we begin the course, it may be the most difficult of questions to answer objectively.
It is relatively easy to answer it subjectively. A recourse United States Supreme Court Justice Potter Stewart famously took when he coined the memorable phrases “I know it when I see it” for defining obscenity in movies during the course of the Jacobellis v. Ohio trial in 1964. His exact quote was as follows:
“I shall not today attempt further to define the kinds of material I understand to be embraced within that shorthand description; and perhaps I could never succeed in intelligibly doing so. But I know it when I see it, and the motion picture involved in this case is not that. – Jacobellis, 378 U.S. at 197 (Stewart, J., concurring).
While scientific objectivity was and still remains all the rage, the ploymath Karl Polyani, amongst others, recognised the innate power of the subjective knowledge each human being posses within their conscientious when he wrote the following:
“I shall reconsider human knowledge by starting from the fact that we can know more than we can tell.” – Polanyi (1966), page 4
Justice Porter’s phrase and Polanyi (1966) acknowledge that we as human beings possess a wealth of subjective knowledge within ourself. The subjective knowledge is not an inferior substitute for the objective knowledge, i.e., knowledge that can be objectively quantified and losslessly communicated. Conversely, subjective knowledge is a complementary source of valuable information.
Subjective knowledge though has its limitations when it comes to defining things as is evident from Justice Stewart’s quote. It is defined from the perspective of the person who is defining things. The problem with subjective knowledge is that while it maybe uncontroversial within communities that thrive on consensus, it is difficult to find consensus across communities.1 One particular community’s consensus on how economic development should be defined may differ from another community’s definition and there is no tractable way of resolving these differences. In defining economic development, one particular community may give more weightage to fighting hunger, while another community in another part of the world give greater weightage to religious freedom or access to medicine and healthcare. It is difficult for an impartial observer to objectively determine whether access to medicine and healthcare is an adequate or inadequate price to pay for losing one’s civil rights.
I have spent inordinate amount of time in deprived areas of the world listening to people. It has left very skeptical of the arm chair philosopher who thinks they know with certainty what is good for others. My study of development is underpinned with healthy hesitancy to over-define the parameters of development. It thus useful to take a brief look at couple of useful ideas that would help us ground this course. The ideas of philosopher Vilfredo Pareto and the paper Dasgupta and Weale (1992) are going to prove to be very useful in this endeavour.
1.1 Pareto Criterion
The eponymous criterion Vilfredo Pareto defined allows us to compare situations. Lets say a society is in situation \(A\). A particular policy \((P_B)\) initiative like upgrading a road that connects cities would put the society the society in situation \(B\). Can we evaluate policy \((P_B)\) compare the situation \(A\) with situation \(B\) and.2
\[\begin{align} {A} \stackrel{P_{B}}{\longrightarrow} {B} \end{align}\]
To compare sitation \(A\) with situation \(B\), we need to find out who the winners and the losers from policy \(P_B\). In a ideal world, if we could identify the winners and losers, we could use to the following Pareto criterion for an improvement: Policy \(P_B\) leads to a Pareto improvement if you can makes some people better off without making anyone worse off. This basically says that for there there to be an Pareto improvement, they should be winners as a result of policy \(P_B\) but no losers. The formal definition of Pareto improvement is as follows:
An intervention that makes some people better off without making anyone worse off is called a Pareto improvement.
If an intervention makes some people better-off and other people worse-off, i.e., it creates winners and losers, then is is not possible to compare the two situation. Any intervention in such a situation is tantamount to redistributing benefits within the society and it is very difficult to find an objective criterion for doing so. Pareto criterion absolves itself from that responsibility. It chooses to confines itself to objective space. In that can both \(A\) and \(B\) are Pareto efficient. The forma definition of Pareto Efficiency is as follows:
Pareto efficiency situations are one where you cannot make anyone better-off without making anyone worse off.
The concept of Pareto improvement and Pareto efficiency underpins almost any study of Economics or Finance. The idea is while you can an intervention that is a Pareto improvement can be objectively said to improve the welfare of the society. These are the low hanging fruits for the society to pick. Once all such intervention opportunities have been exploited, the society has reached all the situations where there are no interventions that can be considered to objectively improve the welfare of the society because while some people “win” others “lose”. To compare the winners and losers, one would have to rely on subjective judgement and these subjective judgements are slippery slope. According to Pareto, once all the low hanging fruits have been picked, the society has reached the state of being Pareto efficient and while redistribution still remains a viable options, there are no options for unqualified gains.
The concept of Pareto efficiency is key to studying Development Economics. It allows mapping the paths along with a less developed society can improve its lot by creating winners without creating losers. Implicit within this is the understanding that society that are less developed forgo opportunities for Pareto improvement due to certain structural constraints and studying development economics can allows us to remove these constraints. It allows development economics to try to search for improvements using objective criterion without getting discussion about whether access to health care is more important than religious rights or vice-versa. These are decisions that societies and communities need to make on their own. While studying them is of great interest, the primary interest in Development economics is to find the low hanging fruits and pick them.
The primary role of Development economics is to find interventions that increase the size of the pie, so that everyone gets a bigger slice of pie than they were getting before the intervention.
The pie here referes to the total output produced by the country or total earned income by the society. If a country’s total low per-capita income is sufficiently growing fast, it is theoretically possible for everyone to be better-off without making anyone worse off.
In reality, it is impossible to imagine a situation where no one is worse-off. If a million people are better-off and 1 person is worse-off, it would not be classified as a Pareto improvement according to the Pareto criterion. Academic has spent a long time trying to find a mechanisms to refine the criterion and make it more operational. Kanbur (2005) is a beautifully written reflective piece on his own struggles with operationalising Pareto criterion in the real world. Ravi Kanbur is a famous Economist who has straddled both worlds of academia and policy making with equal ease and if you want to read further on this topics, Kanbur (2005) is an excellent place to start.
One of the flawed attempts to operationalise this has results in now famous (or infamous for some) cost-benefit analysis. It makes no sense to add different types into one single benefit column and aggregate different types of costs into one single column. That is literally akin to adding up apples and oranges. To get around this problem, what cost-benefit analysis does is to converts all the benefits and losses from an intervention into pecuniary units. Adding up pecuniary values is not a problem. Some benefits and losses are pecuniary to start with and they can be easily aggregated. Other losses and benefits are non-pecuniary. Like the discomfort from noise from the highway or breathing difficulties from due to tyre dust releases due to friction between tyres and asphalt.3 It is obviously difficult to turn all benefits and losses into pecuniary values. What cost-benefits analysis does is either presumes that markets can value each of these costs and benefits. When markets don’t exist, they survey people and ask them to assign pecuniary values to the costs and benefits. Looking behind the curtain of cost-benefit mechanism is literally like looking into a sausage factory. Once you have looked, you never can bear to eat one ever.
1.2 Living Standards and Quality of life
Economics is known as the “dismal science” because its remit is to find consensus where none exists, an ideas first captured in the Arrow’s impossibility theorem Arrow (1951).
Finding tractable policies to address economic development in the absence of universal consensus requires making difficult practical choice. These choices can potentially enrich or impoverish the lives of millions. Any attempt to study economic development starts with choosing to make some practical choice. The Pareto criterion allows to define an area that is uncontroversial without getting lost in weeds of subjective and context-contigent analysis.
In this spirit, we will start by attempting to find certain tractable measures that capture the quality of life as experienced by humans. If I had to hazard a guess, this is easier to do in poorer societies that it is in the richer societies. It is the rich that often quibble interminably with existential navel gazing questions about life, having acquired the means to fulfil the basic necessities. The questions about quality of life are easier to answer when the very basic needs like food, clothing and shelter remained unfulfilled. This is borne out by pattern that emerges from Figure 1.1 below. Figure 1.1 plots life expectancy (vertical axis) versus GDP per-capita (horizontal axis). It shows you how the relationship (black solid line) between life expectancy and GDP per-capita has evolved from 1952 to 2007 and how countries have moved along the relationship. We can also track how some specific countries have moved along the relationship. China and Brazil saw large increases in life expectancy from 1952 to 1972 even though their GDP per-capita did not change significantly during this period. UK, USA, Switzerland and USA saw large increases in GPD per-capita, yet their gains in life expectancy was relatively small. Ethiopia has seen very little change in either life expectancy or GDP per-capita. Botswana is one of the outliers in Figure 1.1. Its life expectancy increases significantly between 1952 and 1972. From 1972 its life expectancy declines quite significantly in spite of a significant increase in GDP per-capita.
Figure 1.1: Life expectancy and GDP per-capita from 1952 to 2007
Figure 1.1 suggests that for countries with low GDP per-capita, small increases in GDP per-capita income can have disproportionately large impacts on life expectancy of its citizens. For countries with high GDP per-capita, life expectancy does not change significantly even if there is a large increase in GDP per-capita. If we were to look at Brazil and China through the lens of Pareto criterion, it does not neatly fits into the Pareto criterion but it could be argued that policies that drove the modest modest increases in per-capita income (i.e., increased the size of the pie) led to large gains in life expectancy were a Pareto improvement (i.e., in a situation where life expectancy increases from mid 40s to mid 70s in a relatively short period of time, there are unlikely to be losers).
Figure 1.2 gives us a more detailed picture of how the life expectancy has evolved in 12 countries across the world. In the top left corner, Figure 1.2 shows the evolution of life expectancy in United States from 1952 to 2007. The life expectancy in United States was 68.44 years in 1952 and 78.24 in 2007. For each subsequent country, the graphs maps how both how the life expectancy evolved from 1952 to 2007 in compare it to the US life expectancy in 1952. Singapore had a life expectancy of 60.4 in 1952 and thus starts below United States and ends with a life expectancy of 79.97 in 2007. The coloured area in the graph indicated how life expectancy in a particular country compares to the United States life expectancy in 1952. High GDP per-capita countries likes Switzerland and United Kingdom look very similar to United States. South Korea, Brazil, China and Guatemala and India significant leaps in life expectancy. It is useful to note that China and India account for one third of the world’s population and if follows that one-third of the world population is living much longer than it used to. It is sobering to note that countries life Botswana, Nigeria and Rwanda have fared less well are exceptions rather than the rule. Crises in Botswana and Rwanda have led to significant drops in their life expectancy and it highlights the limitation of using a simple measure like GDP per-capita.
Figure 1.2: Life expectancy at Birth from 1952 to 2007
While this is far from a perfect example to illustrate the Pareto criterion, it does illustrate the large transformation welfare benefits that can accrue from economic growth for countries with low GDP per-capita income. We follow this example with a description of a seminal paper that still informs policies of development oriented institutions across the world.
Figure 1.3 examines how the living standards have evolved in 12 countries across the world. Each red colouring indicates the distance from United States GDP per-capita in 1950. Singapore, Australia, France, United Kingdom, Spain, South Korea and Botswana have GDP per-capita that is higher than United States GDP per-capita in 1950. China, India, Guatemala and Nigeria have yet to reach the United States GDP per-capita in 1950.
Figure 1.3: GDP per-capita
Dasgupta and Weale (1992) chose six measures to do a cross-country comparison. These measures were per-capita income, life expectancy at birth, infant mortality rate, adult literacy rate, index of political rights and index of civil rights.4 Dasgupta and Weale (1992) rank countries along these six parameters and look for correlation in the rankings. The correlation matrix from Dasgupta and Weale (1992) is reproduced in Table 1.1. They found that life expectancy rank was highly correlated with infant mortality rank and adult literacy rank, but the correlation was lower with index of political rights and index of civil rights. They also found that per-capita income had high correlation with all other variables across the board.
| Life expectancy | Infant Mortality | Literacy rate | Political rights | Civil rights | |
|---|---|---|---|---|---|
| Per-capita income | 0.78 | ||||
| Life expectancy | 0.69 | 0.92 | |||
| Infant Mortality | 0.59 | 0.8 | 0.79 | ||
| Literacy rate | 0.49 | 0.41 | 0.41 | 0.24 | |
| Political rights | 0.51 | 0.43 | 0.38 | 0.27 | 0.79 |
1.3 Aggregate Output and Income
Let’s examine in detail the relationship between life expectancy and per-capita income. We can use use the per-capita gross domestic product (GDP) as the proxy of for per-capita income. The box below explains why we can use GDP per-capita as a proxy for average income or living standards of the people living in the country.
Total aggregate output produced within the country is also the total aggregate income. This may seem counter-intuitive at first but makes sense if you sit back and think about it for a moment.
Every time something is sold in the country, the proceeds from that sale inevitably becomes someone’s income. Broadly, the proceeds from the sale are divided between income earned by the workers, managers and owner of the firm. It follows from basic national income accounting that the total value of the output produced in a period in a closed economy should be equal to the total income earned in a closed economy.
Gross domestic product (GDP) is total output produced and is usually a good proxy for the total income earned in the economy. Divide GDP it by the total population and you get Gross domestic product per-capita (GDP per-capita). It is usually a good proxy for the average income or living standards of people living in the country.
Figure 1.1 show us how us how the relationship between GDP per-capita and life expectancy has changed for countries across the world from 1952 to 2007. What is striking is to the see the gains in life-expectancy in countries like Brazil and China on the lower end of the GDP per-capita spectrum. The gains are much more modest on the higher end of the GDP per-capita spectrum for obvious reasons that are do with life expectancy’s upper limit. This suggests that the greatest benefits of accumulated economic growth are for countries that transition from the lower end of the income spectrum to the middle.
Let’s now view this from a purely practical perspective. Let’s say we had to evaluate the effect of a certain policy, say construction of a highway, on the society. We could choose between using life expectancy and per-capita income as a measure that captures the welfare of the society. While life expectancy is a better choice given that it is highly correlated with all other measures, the problem with life expectancy is that any changes today will be reflected in life expectancy in the next five decades. Conversely, it is much easier to map the impact of a certain policy like building a highway on per-capita income next year. Per-capita income is simply growth of per-capita income accumulated over a long period of time. What this tells us is that if we focus on long-term per-capita income, we are also likely to drive up the other indices of welfare like life expectancy, life expectancy at birth, infant mortality rate, adult literacy rate, index of political and civil rights. The word likely suggests that this is a necessary condition but may not be sufficient condition.
While this is the usual rationale for pursuing economic growth as a means for the end of economic development, what gets lost often is that it is the longer term gains in per-capita income that has the potential of delivering economic development gains. This is often because an increase in per-capita income increases the government’s fiscal capacity, i.e., its capacity to tax the economic activity and spend it effectively in creating public goods. While higher per-capita income creates potential fiscal capacity to provide its constituent population with better public good, it is by no means guaranteed. The country still has to decide whether to tax and provide public good or not to tax. Yet, without an increase in per-capita income, there is no possibility of providing the public good. Life expectancy, adult literacy and infant mortality are directly related to public goods provision.
Our objective in this course is to understand the role infrastructure plays in increasing the long-term per-capita income and increasing a society’s capacity to deliver public goods to its citizens. While it is easy to build schools and hospitals in a developing country, it is more difficult to ensure that teachers and doctors are present in those building to carry out their duties. Banerjee and Duflo (2006) looks at how to address the absence of teachers and doctors and find that if a particular school or hospital is near a good road, the attendance is likely to to be high. Transport infrastructure plays a critical role in reducing the personal transaction cost of teachers and doctors in reaching their place of work and reduces attendance.
Infrastructure plays a significant role in both increasing long term per-capita income of a country enabling it to find resources to provide the public goods as well as reduce the cost of providing the public good.
1.4 What is a production function?
There is two sides to any individual’s materialistic existence in the society. They work to produce things. What they earn from their work, they spend on consumption. Development economics as field has mainly chosen to focus on the production aspect. It has been driven by the conjecture that the problem of development emanates from the production aspects in the economy. Let’s try to answer the following question:
Any production process uses a range of inputs and a set of ideas to produce an output. Think of the production process in a coffee shop. We can broad divide the inputs into three distinct kind of inputs.
- The coffee shop uses physical things like the building, tables, chairs, coffee machines, computers, accounting software etc. These are inanimate inputs and we will put them under a broad category called physical capital.
- The coffee shop requires effort and initiative from a range of workers, i.e., waiters, barista, manager and owner etc. It is useful to note that it is not their mere presence that the coffee shop requires to produce coffee. It is requires effort and initiative from them. We can puts these animate inputs into a broad category called labour.5
- Third and maybe the most important thing that the coffee shop uses is a set of ideas or a plan. The set of ideas or a plan visualise how to use the physical capital and labour to produce the output. Without a functional plan, chaos will ensure and there would never be any hope of ever producing any coffee. The plan is often what distinguishes a coffee shop that you would love to visit from a coffee shop you would go to any lengths to avoid. This is the intellectual input and we will out this under a broad category called technology.
The production function is simply is a relationship that describes how physical capital, labour and technology come together to produce an output at the level of a firm. A firm’s production function is given by the following relationship Using \(K\) for physical capital, \(L\) for labour and \(A\) for technology, we can describe it as follows.
\[\begin{align} Y = F(A,K,L) \end{align}\]
where \(K\) the physical capital, \(L\) the labour and \(A\) the technology used in producing the output \(Y\). This is a very generic relationship and can easily be adapted to describe any production process.
This relationship is very abstract and if every production process had a distinct relationship, it would be very cumbersome to use. Turns out that if the production function exhibits a simple intuitive property of constant returns to scale, it becomes widely applicable.
The idea is simply this. Start with a simple production function \(Y = F(A,K,L)\). What happens to the output when double all your inputs. Let’s say you produce output \(Y'\) if you double all inputs, i.e., \(Y' = F(A,2K,2L)\). See the footnote below for explanation on why \(A\) cannot be doubled.6 We will drop the \(A\) for now an focus on the relationship between \(Y,K\) and \(L\). Once have a beter understanding of this relationship, we incorporate the \(A\) back into our production function in the next chapter. We will proceed with the following production function:
\[\begin{align} Y = F(K,L) \tag{1.1} \end{align}\]
The relationship between \(Y'\) and \(Y\) determines the return to scale. If \(Y'<Y\), then we say the production function has a decreasing returns to scale. If \(Y'>Y\), then we say the production function has a increasing returns to scale. If \(Y'=Y\), then we say the production function has a increasing returns to scale.
The returns to scale can be defined in more generic way. Let \(Y'' = F(\lambda{}K,\lambda{}L)\), i.e., if the inputs were increased by a factor of \(\lambda\), the output would be \(Y''\). In that case the returns to scale can be defined in the following way.
| Returns to scale | Output |
|---|---|
| Decreasing returns to scale | \(Y''<\lambda{Y}\) |
| Constant returns to scale | \(Y''=\lambda{Y}\) |
| Increasing returns to scale | \(Y''>\lambda{Y}\) |
Decreasing returns to scale means that if the production is going to be expanded, the output will not keep pace. Any sensible owner would thus cut back the production. Similarly, increasing returns to scale implies that if the production is going to be expanded, output will increase at a higher rate than the rate at which inputs are increased. In this case, the owner has all the incentive to expand the production process. Constant returns to scale implies that the output will change by the same proportion as the inputs.
A moment’s contemplation should reveal that constant returns to scale says something very useful. It says that scale is not important if the production function has constant returns to scale. It says that if you a firm production process has constant returns to scale, increasing or decreasing the scale at the margin by changing the inputs will not have any impact on the relationship between output and input. The following mathematical derivation captures this intuition. For an arbitrary \(\lambda>0\),
\[\begin{align} Y'' = \lambda{Y} = F(\lambda{}K,\lambda{}L) \end{align}\]
Since this relationship hold for any \(\lambda>0\), we can set \(\lambda=\frac{1}{L}\). In that case \(\lambda{Y} = F(,\lambda{}K,\lambda{}L)\) can be written as \(\frac{Y}{L}=F\left(\frac{K}{L},1\right)\). This can be written as follows:
\[\begin{align} y & = f(k) \tag{1.2} \end{align}\]
where \(y=\frac{Y}{L}\) is the output per worker, \(k=\frac{K}{L}\) is the capital per worker and \(F(\frac{K}{L},1)=f(k)\) is the new transformed production function that describes the relationship between output per worker and capital per worker.
If we were to make the assumption that all the firms in the economy were close to constant returns to scale, that would allow us to use a production function to represent the whole economy. The assumption is not very difficult to justify. If firms owners are ones that optimise their production processes, then they should change their scales in order to be in the constant returns to scale range. That is, if a production process exhibits an increasing returns to scale, they should exploiting the benefits of increasing returns to scale till they reach a point where the production process exhibits constant returns to scale. In the case of coffee shop, that just means that would keep hiring workers and adding equipment till the time the coffee shop reaches constant returns to scale. Further, the owner also not like to increase the scale beyond a threshold where the production process start exhibiting decreasing returns to scale. Hence, it is not unreasonable to assume that all production processes in the economy should be the constant returns to scale range.
The constant returns to scale is an approximation. While there maybe some disadvantages of the approximation in terms of not capturing the intricate details of the reality, there is one huge benefit of using that assumption. It allows us to approximate the production function of the economy. This is because with constant returns to scale, scale does not matter. Hence, whether we divide firms into tiny subdivisions or aggregate them into one larger whole, the relationship between output per worker and capital per worker is not going to change. If the constant returns to scale holds for the most firms in the economy, then we can represent the varied production processes of the economy in on simple relationship. As we will see below, this allows us to look at what has happed to the relationship between output per worker and capital per worker in various economies around the world.
Output per worker
Figure 1.4 graphs the output per worker for United States, Singapore, Australia, France, United Kingdom, Spain, South Korea, Botswana, China, India, Guatemala and Nigeria. While Singapore has outperformed United States in GDP per-capita by a significant margin, it not done so by a significant margin in term of GDP per worker. In terms of worker’s productivity as measured by output per worker, United States outperforms most other countries in the world.
Figure 1.4: Output per worker