A Statistical Argument for Infrastructure Investment

It is well established that a robust infrastructure is associated with the general wealth and well-being of a region. Modern life in a developed region requires reliable roads, railways, ports, airports, and utilities installations for society to have the comforts it desires and to achieve its ambitions.

However, is the association between wealth and good infrastructure just a correlation, or is there a causation as well? In other words, do richer economies have better infrastructure because they already have more resources, or can investing in infrastructure be a lever for a nation to get richer?

In this study, historical data from 47 countries between 1995 and 2021 was collected to help answer this question. The findings were that investing in infrastructure can indeed trigger the enrichment of a nation, but only for non-high-income countries.

Methodology

Let us start by defining the rationale for the analysis and describing our hypothesis. Why would a better transportation infrastructure make a country richer? Note: for this study we will consider being “rich” or “wealthy” as the ability to generate income, a classification applicable for nations and its citizens.

To generate income, people and businesses sell products or provide services, and to increase this income, they must either sell more or charge more (or do some combination of these, sometimes even reducing one to increase the other by a larger factor – the propriety called elasticity). The ability to change the price is limited by the amount of competition (or substitute products) in the market, while the ability to sell more is limited by the reach of a business. It is not always possible to influence the competition, so improving the reach can be the best (and sometimes only) way to grow. For example: a farmer that can grow its produce efficiently and sustainably but lives somewhere with poor physical access to other regions is faded to sell only to its local community. Consumers in other regions would have no choice other than to consume from farmers who are not as efficient as the first one, probably even paying more for a worse product. But if new cost-efficient and sustainable highways, rail lines, or ports were installed, enabling these goods to be transported faster, cheaper, and fresher, the market dynamic would change, favoring the one with the better and more cost-efficient product. Expanding on this example, the producer would have access not only to new consumers but also to new vendors, being able to acquire better and cheaper supplies making their product even more attractive.

There is no shortage of examples for the economic benefits of better infrastructure. But just to ensure we are being comprehensive, let us just consider a few more considering the moving people instead of cargo. Making the movement of people faster and cheaper also present many economic benefits, such as shortening workers’ commute (thus increasing both their productivity and leisure time), increasing the pool of talent that a company can reach, expanding the area that a service worker can attend, or simply enabling more tourists to enjoy a vacation spot.

In summary, infrastructure is the backbone of the developed economic ecosystem. Being able to meet with anyone and ship anything anywhere is what differs today’s way of life from that of the past. Any region without proper installations would be disconnected from the rest of the world (and sometimes even from itself), limiting its growth potential.

?

Data Selection

Two sets of data were required to perform this analysis: how much each country invested in infrastructure and how their overall wealth increased. By making a regression between these variables, we can infer if one affects the other and analyze the statistical significance of this relationship.

The first set of data was collected from the OCDE: total investment in transportation infrastructure over the years by each of their members and some other selected countries as a percentage of their respective GDP for each year. This figure includes both public and private investments in roads, rails, and inland waterways and considers construction of new assets as well as improvements (i.e., repairs and maintenance) to existing ones. It does not include non-transportation infrastructure such as telecommunication towers or energy generation and transmissions installations.

The second set of data was collected from the World Bank Databank: Annual percentage growth rate of GDP per capita based on local currency and adjusted for inflation. Many variables could be used to measure if a nation is better off after certain investments or other types of actions, such as the Human Development Index (HDI). Nevertheless, the growth of GDP per capita was selected because the goal is to find out whether the nation’s population got richer from one year to the next on average. What the nation does with this new income (other than reinvest in infrastructure) is out of scope.

?

Scope of countries

Countries included in this analysis had to satisfy two requisites. First, they had to have data available for at least half of the period analyzed (that is, for at least 13 of the 27 years in scope). Countries with too little data can introduce bias by skewing the result without us seeing how they evolved over time. Second, countries must have a minimum land area. This requisite serves to exclude small, high-density nations or territories such as Monaco, Hong Kong, and Singapore. Even though some of these territories are global benchmarks of infrastructure development and economic progress (even deserving a dedicated study just on them), their sizes make their needs too unique to be included in a study that aims to uncover general insights that can be replicated in other places. For that reason, a threshold was determined in this analysis: territories must have at least 10 thousand square kilometers (3.9 thousand square miles) of land area, which is equivalent to the size of Lebanon or to the state of Connecticut, in the United States.

The resulting list after applying these filters contained 47 countries, which were then segregated by income group based on their classification by the World Bank in Fiscal Year 2023. The countries included were:

• High Income (33): Australia, Austria, Belgium, Canada, Croatia, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Japan, Latvia, Lithuania, Netherlands, New Zealand, Norway, Poland, Portugal, Romania, Slovak Republic, Slovenia, South Korea, Spain, Sweden, Switzerland, United Kingdom, and United States.
• Middle Income (14): Albania, Azerbaijan, Belarus, Bulgaria, China, Georgia, India, Mexico, Moldova, Montenegro, North Macedonia, Russia, Serbia, and Türkiye.

No Lower Income countries passed both criteria to be included in the analysis.

?

Assumptions about the interdependencies between variables

The hypothesis to be tested is if more investment in infrastructure can lead to an increase in the population’s income. In other words, does investing more in this sector get people richer and investing less makes them poorer? Or, in statistical terms, the null hypothesis (the one we want to reject) is that a change in infrastructure investment does not impact the enrichment of a nation.

The challenge in making any association between these variables is that the infrastructure sector typically has exceptionally long payback periods. Building a new road or rail line with all its support installations (e.g., terminals, service stations) can take many years and the return of these investments can even take decades.

To simplify this problem, we will need to make assumptions that, although not perfect, have some logical grounding while enabling an analysis to be performed. The assumption we will make is that most of the benefits of the investments made each year (let us call it year zero) come right in the following year (year one). Note that we are not talking about the financial return to the investor of the project (which as we discussed can take an exceptionally long time), but about the benefit that the local or regional community receives by having that new installation close to them.

We know that having the full benefit practically right away, in the next year, is not necessarily true. Most projects take much longer than one year to be completed and made available to the public. Additionally, even after its completion, the new installation can take even more time to be used at full capacity (thus maximizing its benefits) because people will take some more time to fully adapt to this new option at their disposal.

However, it is not unthinkable to consider this assumption as true, at least partially. First, because the investment of a large project is not made just in its first year: Cash needs to flow in continuously across the life cycle of the project (that is why when financing stops, the project usually halts). Secondly, because the construction itself can already benefit the population by creating jobs and heating up the local economy. In longer, bigger projects it is common to see new settlements forming around the construction site, sometimes even becoming permanent towns. And finally, because the infrastructure investment data used in this study does not include only investments in new projects: it also includes investments in maintaining and improving the existing network. These types of projects are usually less capital intensive than entirely new installations but are much more frequent and take less time, so their benefits can be enjoyed sooner.

For these reasons, assuming that the investments in infrastructure are converted into an increase in the population’s income in the following year can be seen as a plausible assumption while enabling such an analysis to be made.

?

Findings

Curves of investment and income increase

The first thing to do after collecting the data is to get a sense of how both variables behave. In Exhibit 1 we see the investment level in transportation infrastructure for all countries in scope as a percentage of their respective GDP, while in Exhibit 2 we can see the change in each country’s average income (considered as GDP per capita adjusted for inflation).

Exhibit 1 – Investment in Transportation Infrastructure as % of GDP – All Countries

?

Exhibit 2 – Annual change in GDP per capita adjusted for inflation – All Countries

Some readily noticeable observations are that the investment level in the sector being studied is roughly constant. Except for a few outliers, countries usually present a very stable level of investment in infrastructure, changing only a few base points from each year to the next. The average for all countries, represented by the highlighted line, remains slightly above 1% during most of the period in scope, except for some atypical moments such as the 2008-2009 financial crisis. The average investment level for the entire period was 1.16%. The growth in GDP per capita, on the other hand, fluctuated more over time. The effects of the global crisis in the past decade (the 2008-2009 financial crisis and the Covid-19 pandemic in 2020) are clearly visible.

Next, we segregate the data into the two study groups in scope. High income economies directed, on average, 1.00% of their GDP towards infrastructure each year of period analyzed (Exhibit 3), which is lower than the aggregate average of 1.16%. This observation will make sense later in this analysis when we study the expected return on this type of investment. At the same time, their annual income growth stayed around 3-4% during the late 90’s and 2000’s and around 1-2% during the 2010’s (Exhibit 4).

?

Exhibit 3 – Investment in Transportation Infrastructure as % of GDP – High Income Countries

?

Exhibit 4 – Annual change in GDP per capita adjusted for inflation – High Income Countries

For middle income countries, both variables stayed at a higher level than that of high income countries: Average Infrastructure Investment stayed constantly between 1% and 2% (Exhibit 5), averaging at 1.59%, while income growth averaged at 3.98% during the entire period (Exhibit 6).

Exhibit 5 – Investment in Transportation Infrastructure as % of GDP – Middle Income Countries

Exhibit 6 – Annual change in GDP per capita adjusted for inflation – Middle Income Countries

?

Correlation between variables

Having gotten a sense of how the variables in scope behave, we can now analyze the correlation, and hypothesized causation, between them. We will do this by performing a regression analysis where the independent variable (a.k.a. the “lever that we can adjust”) is the infrastructure investment in year X and the dependent variable (a.k.a. the “return of the investment”) is the growth in GDP per capita on year X+1. Each observation in the model is a pair of values that include (i) the investment level for a given country and year and (ii) the growth in average income for that country in the following year. Plotting these observations on a scatter chart and adding a trendline for the linear regression gives us the graph in Exhibit 7.

?

Exhibit 7 – Scatter plot between infrastructure investment and GDP per capita growth – All Countries

?

Exhibit 7 shows that the variables can be very disperse: Virtually all sorts of combinations between infrastructure investment and income growth seem to be possible. Still, a higher concentration is apparent in the highlighted quadrilateral formed by the vertical lines 0% and 2% (level of infrastructure investment) and horizontal lines 0% and 10% (annual GDP per capita growth).

Now, if we focus on the correlation between the variables, we can see that the positive slope of the trendline indicates a positive correlation between them. The values of the trendline (positive slope of 0.639 and constant of 1.89) can be interpreted as follows: On average and considering all other things equal, countries should expect an increase on their population’s income of 1.89% per year plus 0.64% for each percentage of the GDP that is directed towards infrastructure. For instance, for a hypothetical country planning to invest 2% of their GPD towards transportation infrastructure this year, they should expect their GDP per capita to increase in real terms (above inflation) by 1.89% + (2*0.64%) = 3.17% from one year to the next. That means that for every $10,000 in GDP per capita in the first year, the new figure would be $10,317 plus inflation for the period.

Let us look now at how this relationship behaves for the two subsets of the sample analyzed. Starting with the high income countries, we see in Exhibit 8 that the slope of the trendline is much smaller, at 0.16, and the correlation factor R-squared (R2) is almost zero.

Exhibit 8 – Scatter plot between infrastructure investment and GDP per capita growth – High income countries

So far it looks like in high income nations investing in infrastructure is not such a great predictor of income growth as it is for the overall sample. We will discuss this observation further in the next section, but before that let us check the results of the middle-income countries.

The regression analysis for the middle income countries is presented on Exhibit 9. We see here a slope of 0.45 suggesting a stronger relationship between both variables in this sample segment than in the high income group. One important observation is that the trendline constant for this regression is significantly higher than that of the previous two: it is 3.24% for middle income countries while all countries and high income countries had constants of 1.89% and 1.96%, respectively. This higher constant suggests that the combination of all factors that impacted the income growth for these middle income countries other than the variable being studied here – infrastructure investing – makes the expected annual income growth of these countries to be already higher than that of the other group of countries.

Exhibit 9 – Scatter plot between infrastructure investment and GDP per capita growth – Middle income countries

Comparing the trendline constants and the observed average compounded growth in the period for both groups (Exhibit 10), we see that even if most of the growth is explained by other factors in both groups, infrastructure investment was more relevant to the growth of middle income countries (15%) than to the growth of high income countries (9%).

Exhibit 10 – Comparison of income growth components between both sample groups

Statistical considerations?

In a regression model it is important to observe the coefficient of determination R-squared (R2) to see how well the model explains the observed data. The value of this coefficient ranges from 0 to 1 and the higher it is, the better the model fits the observations. The three values or R2 obtained on the three analyses were:

• Regression for all countries: 0.0179
• Regression for high income countries: 0.0005
• Regression for middle income countries: 0.0144

We see that all models had R2 values much closer to 0 than to 1. This propriety is expected considering that we are analyzing economic variables, which are based mostly on human interactions and can be influenced by infinite factors. Expecting that a single variable could accurately predict such a broad indicator would be shortsighted at best. It is not possible to judge if these values are high or low based only on their absolute numbers, but at least it is possible to compare them among themselves: The fact that the coefficient for the high income group is significantly lower than that of the other two models implies a much weaker relationship between the variables for this group.

However, a low fit to the model does not necessarily mean that the model needs to be discarded. It is possible to see if the model is at least statistically significant by analyzing the p-value of these regressions. Basically, considering a 5% confidence interval (which translates to being 95% confident that the model correctly explains the data), the p-value would need to be lower than 0.05 for the conclusion to be statistically significant and the null hypothesis (i.e., that infrastructure investment does not explains income growth) to be rejected in favor of the alternative hypothesis (infrastructure investment can explain income growth). The p-values obtained in the three models were:

• Regression for all countries: 0.0000064 (less than 0.05)
• Regression for high income countries: 0.5303782 (greater than 0.05)
• Regression for middle income countries: 0.0334632 (less than 0.05)

As we can see, the model was able to explain the relationship between infrastructure investment and income growth in the overall model and in the middle income model, but not on the high income model. This finding means that, with a confidence level of 95%, the regression model found for predicting the average income growth of the middle income countries analyzed, based on their respective level of investment in infrastructure, is statistically significant.

?

Study limitations

The greatest limitation of this study was data availability. A lone source of data for each variable was selected to improve the consistency of the observations and reduce the risk of error by handling multiple sources. However, the source for the infrastructure investment data, the OCDE data bank, did not had sufficient or any information on many important markets such as Brazil, Chile, or South Africa, leaving them out of the study. Also, having no information on any low income countries prevented additional and assumingly valuable and interesting insights from being discovered. Information on other types of infrastructure such as telecommunications and energy could have enriched this study even more.

Nonetheless, this study was able to uncover some interesting findings and the limitations found can be addressed in future analyses.

?

Conclusion

In this study we were able to find statistical evidence that investing in infrastructure can, indeed, enrich a nation, but only for middle income countries. This finding is not surprising given that middle income economies are greatly dependent on extraction and manufacturing activities, which require a robust transportation network to move materials and products, while the economy of more developed nations, even those with significant industrial footprint, rely more on services than the nations with lower income.

But does this finding mean that wealthier economies should stop investing in infrastructure altogether? Absolutely not. First, because in the hypothesis tested, we were only unable to prove that a relationship between the variables existed for these countries, but it does not mean that such a relationship does not exist. Second, because even if infrastructure investment were completely unrelated to income growth in the short term, it does not mean that it would not impact the nation’s income in the long term, specifically considering that neglected transportation assets inevitably cripple the economy and lifestyle of its citizens after some time. And third, because we analyzed the returns of infrastructure investments only through economic lenses, while countries have multiple other critical interests dependent on infrastructure to consider, such as national security and mobility independence.

At this point, one might also be asking “Ok, investing in infrastructure is good, at least for some countries. Why aren’t these economies investing even more then? Shouldn’t they go all-in?”. Even though through this study we understood better the relationship between infrastructure and economic growth, countries should be careful as to exactly what and how to invest in. History shows that governments that invested too much and too fast in infrastructure (and other areas) in the last century ended up with severe and resilient financial crises due to quick debt increase and super inflation. To avoid such risks in these types of investments, the below actions are recommended:

1. Identify a real demand. It can be easy to fall into the lure of the “build and they will come” mentality. Even though such types of projects can create some local demand just by existing, to be long-lived the project must address an existing, latent demand. For instance, regions that extract raw materials for industries that need to be far away and are connected only by roads might be a great candidate for a new rail line, especially if the supply relationship is expected to last long.
2. If there is no demand, create it (or coordinate with others than can). Imagine a nation that wishes to develop a region that has a shortage of proper installations. Even building a simple, decent quality road that leads there could take a long time to pay for itself, that if it ever does. Developing entire regions is a difficult endeavor, but not an impossible one. A general rule is that assets must be used as much as possible to maximize their returns. So, if there is a plan to build a better road for a distant community, for instance, the responsible for the project should consider if the new use of the road would justify its creation. If not, the project undertaker can assess whether there are untapped economic opportunities in that region before abandoning the project, even if those opportunities are in sectors outside their field of experience. For example, some remote regions could have specialty natural resources important for some industries while others could be a potential exotic tourism hotspot. If the expected usage of a project is below the minimum limit to make it viable, cross-industry coordination could help each part to benefit from this initiative. Private-public partnerships (PPPs) are an example of such type of coordination. Such collections of different but complementary projects can be called a program and would require exemplary governance practices to minimize the risks of one party not being able to deliver their part, bringing losses to the others. Some of these practices are appointing a program overseer and establishing conditional agreements to ensure that the interests are aligned throughout the program lifetime. It is a high-risk and demanding initiative, but one that can bear great returns if done properly.
3. Define the funding. The funding of private-sector infrastructure investing is very straightforward: The rate of return of an investment must be higher than the cost of the capital used to fund it, not mattering if that comes either from debt issued or from equity sold. For example, if a mining company can prove that expanding a rail line and a port that they own will increase their sales, funding these projects should not be a challenge. And with the spread of the ESG agenda across multiple sectors, benefits other than financial gains began to justify infrastructure investments, such as fleet electrification, development of local communities, or sale of carbon credits. For public projects (that is, those made by governments), the process sometimes is not as simple, especially because governments have many more interests to manage simultaneously. They do have the interest to develop regions economically, connect different communities, and improve the lives of its citizens. But when initiatives to tackle these issues cost unjustifiable amounts of money, larger problems may appear in the entire region attended by that government, such as devaluing the currency for everyone that uses it. Governments should be as rigorous as private organizations, if not more, in managing these types of projects. Sometimes they can do a better job representing their citizens by focusing their efforts in designing better policies, choosing the right incentives, and enforcing best practice laws than by directly managing the projects.
4. Avoid inefficiencies, always. Inefficiencies can take many forms, such as poor project management or corruption of officials (on projects where governmental bodies are involved). Unfortunately, until today and even in very developed economies, it is common to see project managers taking unnecessary and uncalculated risks to speed up a large project or to watch scandals take the headlines, halting projects important for the public. To help to implement and take the most advantage of “physical infrastructure” projects for the betterment of a community, it is important to also implement and maintain what is called an efficient “institutional infrastructure”. This concept is composed of a series of processes aimed at identifying, correcting, and preventing bad practices to deviate a good project from its expected positive outcome. Measuring progress, periodically assessing risks, updating expected returns based on new information regarding the demand for that project, enforcing and updating applicable laws and regulations, changing project leadership when required, having a compliance mechanism than can correct issues and prevent new ones, and having unbiased and unconflicted legal institutions and watchdog agencies looking over such projects are some examples of what an environment with a mature institutional infrastructure looks like.

The path to the enrichment of a nation is complex, where many elements are at play, some of which can be more easily influenced than others. This study analyzed the effects of one of these elements and concluded that it can, at least for certain types of nations, help them to achieve their ambitions, even though it is not a risk-free process.

Note from the author

If you enjoyed this article or would like to discuss this topic with me, please feel free to reach out. I am always happy to discuss ideas for new studies that may uncover interesting and valuable new insights, either for the macro economy as a whole or for some specific industries or businesses.

Data Sources

• OECD (2023), Infrastructure investment (indicator). doi: 10.1787/b06ce3ad-en (Accessed on 20 July 2023).
• World Bank national accounts data, and OECD National Accounts data files. DataBank ID: NY.GDP.PCAP.KD.ZG.

About the author

Adriano Claudio is a management consultant with over 13 years of global experience in the infrastructure and manufacturing sectors, among others, having delivered over 25 high-impact business strategy and supply chain management projects for some of the world's largest organizations. Between his projects, Adriano researches topics of his interest applying techniques developed throughout his career.