Detailed files are available above. This page focuses on reporting results and commenting on the visuals associated with the project.
Prior to approximately 1760, the world economy was almost exclusively agriculture-based. Beginning in the late 1700s, and throughout the 19th century, the Industrial Revolution transformed the economies of developed and developing nations. Industry is broadly defined as the production of goods and related services within an economy. It is typified by manufacturing, often on an assembly line. Ultimately, industry-related economic growth peaked in Europe and the United States in the early 20th century.
More recently, the world’s industrialized economies began a transition to “post-industrial,” or service-based, economies. The precise starting date of this transition is difficult to trace. An aim of this analysis is to gain a fuller understanding of when this transition began, whether it has peaked, how it is proceeding now. The analysis will also consider which countries have the most vibrant service economies.
The United Nations maintains a classification system known as the International Standard Industrial Classification of all Economic Activities (called “ISIC”). This schema neatly groups economic activities into various “divisions.” A few pertinent examples of these divisions are:
- 01 - Crop and animal production, hunting and related service activities
- 30 - Manufacture of other transport equipment
- 33 - Repair and installation of machinery and equipment
- 62 - Computer programming, consultancy and related activities
- 63 - Information service activities
These divisions are aggregated into three larger groups called sectors. These sectors are named after the economic epochs previously discussed: Agricultural, Industry, and Services. The United Nations defines these sectors in terms of the groups that they aggregate, specifically:
- Agriculture - ISIC divisions 1-5
- Industry - ISIC divisions 10-45
- Services - ISIC divisions 50-99
The divisions (a two digit schema) are themselves aggregates of industry classes, which are a four digit scheme with the first two coming from the division into which it is classified. For example, Division 25 is described as “Manufacture of fabricated metal products, except machinery and equipment,” and it contains occupation classes 2511 (“Manufacture of structural metal products”) and 2512 (“Manufacture of tanks, reservoirs and containers of metal”).
This analysis will address the following questions:
- Has service-sector economic activity reached a peak? Or is the service economy still growing?
- Does service-sector economic activity tend to be concentrated within any particular geographic regions?
- Does service-sector economic activity tend to be concentrated within any particular economic groups? (ex: high income countries, low income countries…)
- Do countries tend to progress through an industrial revolution before developing a significant services sector?
This analysis has particular meaning for me.
My undergraduate education was Mechanical Engineering, and recently I completed my MBA. My plan is to leverage my business knowledge, analytical abilities, and the content of this nanodegree to transition into a Data Science career.
My prior employment has been in the Los Angeles aerospace and central valley agriculture industries. The applicable ISIC industry classes for these jobs are 3030 (“Manufacture of Air and Spacecraft and Related Machinery”) for the former, and 3320 (“Installation of Industrial Machinery and Equipment”) for the latter. These classes fall squarely within the Industry sector (10-45).
The activities of a Data Analyst appear to fall under either industry code 6209 (“Other Information Technology and Computer Service activities”) or 6311 (“Data Processing, Hosting, and Related Activities”). The applicable division is either 62 (“Computer programming, consultancy and related activities”) or 63 (“Information service activities”). This places data science work within the Services sector (50-99). Thus, my present career transition amounts to a transition from working in the “industry” sector to the “service” sector.
A major motivating factor for my transition into data science is my intention to capitalize on future employment trends. Research I conducted while pursuing my MBA confirmed the common perception that some of today’s best opportunities for salary and career growth are in the computer science and data science fields. Computer and data science occupations are obviously a tiny part of the Service sector, but a purposes of this analysis is to determine whether the service sector as a whole can be expected to continue growing as rapidly as it has been.
Just as important as compensation and growth opportunities, service-sector employees I have encountered throughout my career have tended to be more competitive and higher-performing than those I have encountered working within industry and agriculture. I believe this is because higher-growth (and therefore more lucrative) service-industry jobs attract high-quality talent.
For these reasons, I believe my current career transition will ultimately pay off, both financially and in terms of work satisfaction.
This analysis makes use of four datasets provided by WorldBank.org. The datasets utilized were downloaded from WorldBank.org on January 25, 2018. The specific URLs are as follows:
Agriculture (% of GDP) - http://data.worldbank.org/indicator/NV.AGR.TOTL.ZS
Industry (% of GDP) - http://data.worldbank.org/indicator/NV.IND.TOTL.ZS
Services (% of GDP) - http://data.worldbank.org/indicator/NV.SRV.TETC.ZS
GDP/capita (US dollars, inflation-adjusted) - http://data.worldbank.org/indicator/NY.GDP.PCAP.KD
For details on cleaning and manipulation of the raw data, see the Jupyter notebook for this project.
Exploratory Data Analysis
1. Has service-sector economic activity reached a peak? Or is the service economy still growing?
This question will be addressed for the United States, Europe & Central Asia, and the World at large.
The following figure shows the percentage of GDP contributed by each economic sector for each of these geographic regions. As shown, the service economy has grown in terms of percent GDP.
Agriculture percent contribution has remained relatively constant, especially for the United States. The industry sector has therefore borne most of the relative decrease to “make room” for the expanding service sector.
This growth of the service economy is even more dramatic when viewed on the basis of per-capita GDP for the US, Europe & Central Asia, and the World. See the following.
Perhaps the most striking way of describing the impact of emerging service economy is as a percentage of per-capita economic growth between 1997 and 2015. The following cells present this breakdown for the United States, Europe & Central Asia, and the World.
As shown above, growth in the service sector accounts for 94% of US per-capita GDP growth from 1997 to 2015. It accounted for 87% of worldwide GDP growth over the same time period. Thus, for the time being, the service sector appears to be the dominant driving force for continued national and global economic expansion.
On the basis of the trajectory of the plots above, and the foregoing analysis, the service economy can be expected to continue growing, both in in terms of percentage of global GDP and in absolute terms.
2. Does service-sector economic activity tend to be concentrated within any particular geographic regions?
As shown in the plots below, in 2015, the service sector accounted for a majority of the economic output for all of the regions with available data.
North America has the largest service sector on a percentage basis, at 79% of total regional economic output. South Asia has the smallest service sector, at 54% of total output.
Closer inspection reveals that the regions with the most developed economies also tend to have the largest service sector economies.
3. Does service-sector economic activity tend to be concentrated within any particular economic groupings of countries? (ex: high income countries, low income countries…)
As shown below, for any given year, higher income countries have larger service sector economies. Lower income countries have larger agricultural sector economies. The countries with moderate incomes tend to have the largest concentrations of industry.
The trends are very clear, and can be accounted for by considering how developed the economies within each grouping are. High income economies can be expected to be most developed, and therefore have passed through both economic revolutions (industrial and service). Low income economies are least developed, and therefore a large part of their economic activity is still attributable to agriculture.
It is also helpful to visualize the changing economic constituencies of the different economic groups over time. The data is visualized at 9-year intervals over the time period for which data for all four income groups are available (1997 to 2015). Visualized in this manner, each group’s economic progression over time becomes readily apparent.
Note that for each 9-year period visualized, agriculture’s contribution to each region’s economy decreases. Services’ relative contribution, on the other hand, increases for each country over each time period. For some time periods, industry’s contribution to the lower income group increases. This indicates that some lower income countries may be going through a delayed industrial revolution.
As shown in the table lookup produced below, the countries with the smallest service sector economies tend to be low income. The most common regions for low income countries was Sub-Saharan Africa. Conversely, countries with the largest service sector economies tend to be high income. These high income countries are primarily advanced economies in Europe and Central Asia.
4. Do countries tend to progress through an industrial revolution before developing a significant services sector?
China (see chart below) exhibits the exact economic progression that one would expect, given what economists have come to understand about economic development. In particular, the expected pattern is that agriculture begins as the dominant economic sector, but it declines in relative importance as industry, and later, services, increase.
After some examination of development patterns for various countries, it appears China may be atypical for its very large industry base. China is frequently compared to Brazil, Russia, and India since they are all rapidly developing economies (and known collectively as the “BRIC” countries). For each of those three countries, the services sector has been dominant for a number of decades, and eclipsed industry earlier than might be expected.
India’s pattern of economic development, where services become the predominant sector and industry remains relatively constant over time, appears to more typical of economic development of countries than China’s massive industrialization.
See three further examples superimposed on below. The countries are dissimilar with regard to geographic region classification, so these development patterns are not peculiar to a particular geography or set of natural resources. Rather this pattern appears to be fairly consistent for developing nations.
Finally, an examination of the economic breakdown of countries classified by the UN as “low income” shows that the average impoverished country may not heavily industrialize at all. Rather, it appears they often emerge from agriculturally-based economies straight into predominantly service-based economies.
Several conclusions can be presented from the analysis presented herein. Among these:
- The service economy has been and will most likely continue to be the primary engine for economic growth, both nationally and globally, for the foreseeable future. The previous two decades have shown substantially increased economic output driven primarily by the service sector.
- Service sector economic activity is broadly distributed with regard to geographic region. In 2015, no geographic region profiled herein had a service sector that accounted for less than 50% of its economic output.
- Service sector economic activity forms a larger part of higher-income, advanced economies than it does for lower-income, less advanced economies. With this said, the difference is not so pronounced as one might expect. In 2015, service sectors accounted for 48% of total economic output for low income countries. In the same year, the service sector accounted for 74% of total economic output for high income countries.
- It is not necessarily true that a large industrial sector always proceeds development of a large service sector. In fact, this analysis presented 4 examples of countries wherein the service sector became the largest part of the national economy before industry had even surpassed agriculture as measured by percent of the overall national economy. Despite this, as countries develop and increase in per-capita income, the tendency is for the service sector to gradually eclipse the agriculture and industry sectors. In the same way that Africa has been able to adopt cell phone technology without the intervening landline technology, today’s advanced economies likely “paved the way” for today’s developing economies to establish robust service sectors without needing to make substantial investments in industry.
Possible Improvements to this study
- Perform country-by-country analysis of the order of emergence of industry and service sectors.
- Obtain a longer-term dataset. As discussed in the introduction, the industrial revolution began long before 1960, the start date for this dataset.
- Obtain a more complete dataset. Only 12% of the countries have complete data from 1960 to 2016. For the 25-year time period from 1990 to 2015, roughly 50% of countries have complete data.
- Perform more robust error checking and handling. Another way to handle percentages that sum to something other than 100% would be to introduce a scaled weighting factor that would scale each sectors’ contribution such that they together sum to 100%. Whether or not this is appropriate depends on the nature of the problem with the data.
Analytics Lessons Learned
- Matplotlib allows for incredible precision in generating visualizations, because everything is performed programmatically.
- In my opinion, however, what you gain in visualization precision (as compared to a dedicated tool like Tableau) is not necessarily worth the extra time to generate them. The visualizations generated as part of my analytics jobs project appear much more polished than these, and the effort to generate them was much less.
- The best tool I’ve used for data cleaning is still Pandas and Python.