FOCUSED AND QUICK (FAQ) Issue 166
A STUDY OF DEINDUSTRIALISATION
AND ITS IMPLICATIONS ON THAILAND’S PRIVATE INVESTMENT
1 May 2020
“An examination of literature and data suggests that Thailand
may be on the road to an unhealthy deindustrialisation”
In recent years, Thailand has seen a stagnated private investment. One potential explanation is deindustrialisation – a process commonly defined as a sustained decline in manufacturing share in the economy. To be clear, deindustrialisation itself is not necessarily undesirable, as long as the country manages to reap benefits during the industrialised period and makes the necessary structural transformation to sustain productivity.
In the case of Thailand, however, the economy started to deindustrialise while its labour productivity remains low and the country is already entering an aging society. If such underlying fragility is not yet addressed, our deindustrialisation process is prone to be unhealthy. In order to sustain economic growth in the longer term, the government must prioritise structural policies in two main areas, by undertaking infrastructure investment in strategic areas and by revisiting competition policies in order to encourage firms to innovate and invest.
In recent years, Thailand
has seen a stagnated private investment. Figure 1 illustrates the
falling share of nominal private investment in GDP from 2012 onwards. While a
number of reasons, both cyclical and structural, can be responsible for the
stagnation, this article takes a closer look
at a structural cause – deindustrialisation, the shrinking share of
manufacturing sector in the economy.
In a way, deindustrialisation can be seen as inevitable. It has been taking place in developed countries since 1950s, and is beginning to unfold in developing countries, including Malaysia and Indonesia . Such global phenomenon matters as it has direct implications on growth. Structuralists often see manufacturing sector as a major growth engine in the economy due to its high potential for economies of scale and innovation. Deindustrialisation is thus associated with a decline in technological dynamics, as well as a loss of good jobs and rising inequality (Rodrik, 2015). Therefore, a key question that policymakers need to ask themselves is whether the process of deindustrialisation in their countries is a healthy one – where productivity and growth can be sustained after the industrialised period. In the early stage of deindustrialisation, it is also worth asking if the country has prepared policies to facilitate a smooth transition process.
 The Edge Financial Daily, on August 27, 2018 and Tempo, on February 09, 2017
Figure 1 Share of Private Investment in Thailand’s GDP
Source: National Economic and Social Development Council (NESDC)
This article is arranged in the following orders: the first section starts with a discussion on the definition of deindustrialisation and shows that Thailand is in the early days of the process. Section 2 then discusses whether deindustrialisation in Thailand is healthy or not. Section 3 relates deindustrialisation to private investment, using panel data and fixed effect regression. Lastly, Section 4 offers some policy recommendations.
1. Definition of deindustrialisation
Deindustrialisation most commonly refers to a fall in the share of manufacturing in total employment. It has also been defined as a sustained fall in the share of manufacturing in both total employment and GDP. As demonstrated in Figure 2, Thailand is experiencing a small but noticeable decline in the share of manufacturing in total employment. However, in terms of income share, labour seems to reap less returns from Thailand’s economic growth while capitalists have been capturing higher returns. Moreover, the share of manufacturing in GDP has dropped markedly, after having peaked in 2010 (Figure 3).
Figure 2 Share of manufacturing in Thailand’s total employment
Source: Labour Force Survey (LFS)
Figure 3 Share of manufacturing in Thailand’s total GDP
Should policymakers be alarmed by such statistics? While structuralist approach highlights the importance of manufacturing sector as a growth engine, there are mixed and inconclusive empirical findings on the impact of deindustrialisation on economic growth.  The effect of deindustrialization seems to vary, depending on its causes. Accordingly, this article is going to examine each of the factors behind deindustrialisation.
 Tregenna, F. (2015). Deindustrialisation, structural change and sustainable economic growth, MERIT Working Papers 032, United Nations University - Maastricht Economic and Social Research Institute on Innovation and Technology (MERIT).
deindustrialisation process in Thailand benign or detrimental?
This section explores common causes of deindustrialisation, and attempts to map them with empirical evidence in the case of Thailand. The ultimate goal is to determine whether deindustrialisation process in Thailand thus far is benign or detrimental.
Deindustrialisation is a normal process of development. It is argued that there exists an inverted-U shape relationship between the share of manufacturing employment and income, as shown in Figure 4 . During the dawn of development, the rise in income allows consumers to spend more on basic goods, and later extend their spending on more manufactured durables, such as furniture, automobiles, etc. The share of manufacturing in total employment and GDP therefore rises to meet this demand. At some point, everyone owns the products and further production rests upon replacement spending, which is relatively small. This development encourages consumers in countries with higher income to spend more on services, and thus explains the falling share of manufacturing employment (Lawrence and Edwards, 2013). In short, deindustrialisation is a natural process that occurs to every country after reaching a certain stage of development.
 A similar diagram can be obtained using share of industry to GDP; however, the R-squared is smaller at 0.09.
Figure 4 Inverted-U shape relationship between employment share and income
Note: The y-axis represents the log of
industry share of total employment (industry comprises manufacturing,
utilities, mining, and construction sector). Each dot represents countries at
different stages of development. R-Squared = 0.41.
Source: World Bank (2017)
A more important question to ask is whether the country is ready to deindustrialise. In the case of Thailand, the answer is unfortunately no. As seen in Figure 4, Thailand remains caught in the middle-income trap, with GDP per capita (PPP constant 2011 international $) at $16,286 as marked by a yellow dot in the diagram. The country has not yet reached the pinnacle of the inverted-U graph. This means Thailand has not attained the level of income associated with the ‘natural’ deindustrialisation (approximately $33,240, marked by the grey bar). In technical terms, Thailand is experiencing premature deindustrialisation.
2.1 The productivity argument : better technology replacing labour in the manufacturing sector.
One key feature of the manufacturing sector is its high potential for productivity growth, compared with other sectors. This is often made possible by the advancement of technology, particularly factory automation. As such labour-replacing technology brings about higher productivity, manufacturing employment falls, while its contribution to GDP remains the same – hence a deindustrialized process. 
Despite a rise in productivity, it could be the case that a country still has relatively lower productivity compared to other countries. This has in fact become more prominent with the speed of globalisation. As seen in Figure 5, despite a rise in productivity, Thailand still has substantially lower level of productivity and lower productivity growth than those observed in developed countries and some developing economies. Foreign firms therefore find it more attractive to expand investment outside Thailand, or relocate existing plants to countries with lower costs of production – gradually causing deindustrialisation in Thailand.
In short, despite a rise in productivity, a country can still deindustrialise prematurely if it fails to catch up with its more productive peers. Deindustrialisation in this sense can be considered unhealthy.
Figure 5 Labour productivity in developed and developing countries
Source: World Bank
 In fact, Rodrik (2015) developed a model arguing the opposite – that in small open economies, the rise in productivity should lead to industrialisation, not deindustrialisation. The explanation is that producers in small open economies are typically price takers in the world market. Higher productivity means they can still sell at the same price and earn more profits. Therefore, it makes sense for them to produce more manufactured goods – thereby speeding industrialisation, both in terms of employment and output. Recently, however, we observe a rise in productivity and deindustrialisation occurring at the same time in developing countries Therefore, it is likely that Rodrik's model leaves out certain factors that could explain such deindustrialisation phenomenon in developing countries.
2.2 The market structure argument
In the case of Thailand, the reason for low relative productivity can be attributed to its market structure that has become more concentrated over the past decade. According to Apaitan et al. (2019), the dynamism of Thai businesses has been decreasing while the market power of Thai firms has been increasing over the past decade. In addition, they find that market power is negatively associated with firm's investment, propensity to export, diversification of export products, and likelihood of product upgrade. It also has a non-linear relationship with productivity growth. In this sense, it can be argued that Thailand is not experiencing a benign deindustrialisation, and the concentration of market power might be part of the problem.
2.3 The ‘Dutch Disease’: a specific conceptualisation of the Dutch disease .
Palma (2014) proposes that economies with trade surplus in manufacturing may face deindustrialisation in response to the following shocks;
1) Discovery of natural resources
2) Boom of service exports
3) Change in macroeconomic Policy
To illustrate, economies such as Switzerland, Hong Kong, and Greece deindustrialised because of flourishing exports of financial and tourism-related services. Others such as those in Latin America experienced a sudden switch of policy, from import-substituting industrialisation supported by the public sector to a trade and capital liberalisation, followed by a collapse of the manufacturing sector and the employment structure reverting back to those with trade surplus in primary commodity.
In the case of Thailand, the tourism sector has seen unprecedented growth, in line with a substantial growth of Thailand’s net travel receipt and a substantial increase in the share of services in Thailand’s GDP. However there seems to be inconclusive evidence of the Dutch disease in Thailand, if measured in terms of employment share. On one hand, Chantapong et al. (2015) points out that there is very little flow of labour between manufacturing and tourism-related sector, meaning that, the tourism boom thus far has not drawn human resources from the manufacturing sector in a significant way. On the other hand, we start to observe a higher proportion of new labour joining the booming hospitality industry, as the gradual rise in wages offered in the sector makes it more attractive.
Although there is not yet conclusive evidence of the Dutch Disease occurring in Thailand’s tourism sector, the development in other aspects thus far does not seem healthy. For example, the tourism-related sectors still rely heavily on the attractiveness of the country’s natural sights, which induce relatively low tourist spending and therefore its contribution to GDP remains low. From the labour market perspective, the sector mainly employs low-skilled labour and therefore the wages offered are still far below those in the manufacturing sector. In short, the growing services sector in Thailand has not contributed enough to the overall growth to offset the potential decline in the manufacturing sector. This is because the majority remains traditional, relatively low value-add services.
 The term Dutch disease was first coined by The Economist (1977), used to describe how the discovery of gas reserves in the Netherlands led to lower investment and higher unemployment rate. Later, the meaning evolved to include any negative economic impact stemming from the discovery of natural resources. Palma (2014) proposed his concept of the Dutch disease to describe how economies deindustrialise due to three types of shock, including the ‘original’ cause - the natural resource boom.
2.4 Other explanations
Apart from the causes of deindustrialisation which are most commonly referred to in the literature mentioned above, there are also other plausible explanations – mostly related to the changing business and economic structure. One explanation is that some activities, such as cleaning or security, which were originally done in house in the manufacturing sector, are now outsourced to firms classified as part of the service sector. Therefore, the observed signs of deindustrialisation, in terms of employment share, can be considered a mere statistical illusion.
Another explanation which may be more relevant to Thailand is the fragility of domestic demand for manufactured goods. This is, in fact, a reflection of the aforementioned low labour productivity, which in turn leads to low wage growth and thus subdued consumption growth. Going forward, there is also additional pressure from ageing society, resulting in the lack of firm’s incentive to expand production capacity for consumer goods.
In summary, an examination of the causes of deindustrialisation in Thailand suggests that a combination of factors – both push and pull – are at play, including comparatively low labour productivity and domestic demand. The next section will explore the implication of this phenomenon on investment dynamics of the country.
3. Quantitative study on the relationship between deindustrialisation and investment
The previous sections present several evidence suggesting that Thailand may be on the road to an undesirable deindustrialisation, as the country is still trapped in the middle-income bracket and its labour has seen somewhat stagnated productivity even before entering a fully aged society. This section attempts to analyse these factors in a more systematic way, and in particular, to establish a relationship between deindustrialisation and investment, which used to be the key growth driver of GDP . This will then lead to the final section which aims to suggest ways to reinvigorate domestic investment while deindustrialisation is still in its early stage
3.1 Literature Review
While there are genuine concerns about how deindustrialisation may have a negative impact on investment, there are very few empirical studies  that try to quantify such effect. Two studies seek to establish a relationship between investment, growth, and employment – but in the sense of how much investment translates to the manufacturing share in GDP and employment.
Firstly, Garcia-Santana et al. (2016), using panel data, demonstrated that the manufacturing sector consumes more investment goods as shown in Figure 8. However, they used investment rate as an independent variable and the industry share in GDP as a dependent variable, finding that a 1% increase in investment rate is associated with a 0.55% increase in the contribution of manufacturing in GDP.
Rowthorn and Coutts (2013) also used investment rate as an independent variable and manufacturing employment share as a dependent variable, concluding that a 1% increase in investment rate is associated with 0.33-0.35% increase in manufacturing employment share, depending on different specifications. All three studies employ panel data and fixed effect.
Figure 8 Sectoral shares for different goods (panel data)
Sectoral shares is calculated using data from WIOD at current prices, controlling for level and square of log GDP per capita and country fixed effect
Source: Garcia-Santana et al. (2016)
 In terms of investment rate, defined as a percentage of Gross Fixed Capital Formation to GDP.
 One study which looks at factors affecting investment (and saving) rate is Park and Shin (2009). However, it left out the deindustrialisation-related variable which is of interest of this article.
3.2 Methodology and Data
In line with existing works, this study employs panel data but focuses on the deindustrialisation question. Panel data is used for two reasons. Firstly, the availability of time series data on Thailand’s employment share is limited and neither simple lead-lag test nor Granger causality test showed a sign of relationship between employment share and investment. Secondly, the use of panel data allows control for time and country fixed effects, offering a more robust regression result. The choice of fixed effect panel regression is also supported by Hausman test.
For the baseline result, the dependent variable is investment rate (percentage share of Gross Fixed Capital Formation to GDP), and the independent variable of interest is the percentage share of employment in industry (deindustrialisation variable).
Control Variables are GDP growth and lagged GDP growth to account for how firm’s profitability determines investment and employment decisions, and (log of) GDP per capita and GDP per capita squared to allow for different investment and employment structures of countries at different stages of development. Other variables capturing the structure of the economy include old age dependency ratio  and trade openness  to accommodate the fact that demographic factors affect employment in industry. For instance, older generations are more likely to leave manufacturing sector as they cannot catch up with new production technology. Moreover, openness to greater foreign trade is associated with higher investment and job creation in the manufacturing sector. Productivity variable  is also added to account for the productivity argument mentioned in Section 2. All data is obtained from the World Bank website, covering 164 regions from 1991 to 2017.
 Share of people older than 64 to the working-age population (age 15-64)
 Sum of merchandise exports and imports as percentage of GDP
 Value added per worker in industry
Table 1 exhibits the baseline result with region and time fixed effect. The coefficient of deindustrialisation variable is statistically significant, implying that a 1% increase in the number of manufacturing workers is associated with a 0.35% increase in investment rate. Most of other coefficients have expected signs such as GDP and lagged GDP growth. The humpback shape of the investment rate function (plotted against income per capita) is also evident. Aged dependency rate has a negative effect on investment rate as expected, but the coefficient is not statistically significant. Greater openness to trade on average, does not translate to higher investment rate. The coefficient of productivity variable has an unexpected negative sign but is small and not statistically significant. A more thorough study on the impact of productivity may require a variable representing relative productivity or even an instrumental variable for it to take into account for a two-way causality.
Table 1 Baseline result 
Note: The values in parenthesis are standard deviations.
*, **, *** denote significance at 10%, 5% and 1%, respectively.
 As noted in the literature review, there is a potential endogeneity problem (investment rate affects employment in manufacturing and vice versa). A Granger causality test suggests there is indeed such problem. One solution would be to use an instrument variable that only affects investment rate through employment in manufacturing, opening a room for further study.
An examination of the causes of deindustrialisation in Thailand suggests that a combination of factors is at play, most of which are structural in nature, particularly low growth of labour productivity and demographic change. In fact, both factors reflect the underlying fragility in Thailand’s domestic demand, and deindustrialisation is simply a symptom, reflecting that these structural problems have not yet been addressed, and therefore discouraged firms from investing domestically. The empirical result confirms such phenomenon by showing how the shrinking manufacturing sector is associated with lower investment rate.
Going forward, a transition towards a fully aged society and rapid technological changes represent additional challenges which could further dampen the prospect of domestic demand. In order to ensure a healthy deindustrialisation and sustain growth in the longer term, the government must prioritise policies that strengthen domestic demand, which would subsequently encourage firms to invest more domestically. For instance, the government should undertake infrastructure investment in strategic areas, such as digital technology and human capital, in order to prepare its labour force for the new competitive landscape. Secondly, it should revisit the policies that enhance the degree of competition in the market. This would not only boost the overall investment, but also facilitate innovation and capacity development of smaller firms. By undertaking structural reforms in these areas, Thailand should be on the road towards a healthy deindustrialisation, where growth is more sustainable and inclusive.
Apaitan, T., Banternghansa, C., Paweenawat, A., and Samphantharak, K. (2019). Towards a Competitive Thailand: The Role of Market Power and Business Dynamism. Puey Ungphakorn Institute for Economic Research, Bank of Thailand.
The Economist. (2014). What Dutch Disease Is, And Why It's Bad. [online] Available at: <https://www.economist.com/the-economist-explains/2014/11/05/what-dutch-disease-is-and-why-its-bad> [Accessed 10 April 2020].
García-Santana, M., Pijoan-Mas, J., and Villacorta, L. (2016). Investment Demand and Structural Change. London, Centre for Economic Policy Research
Lawrence, R. and Edwards, L. (2013). US Employment Deindustrialization: Insights from History and the International Experience. [online] Available at: https://www.piie.com/publications/policy-briefs/us-employment-deindustrialization-insights-history-and-international [Accessed 4 Nov. 2019].
Palma, J.G. (2014). De-industrialisation,‘premature’de-industrialisation and the dutch-disease. Revista NECAT-Revista do Núcleo de Estudos de Economia Catarinense, 3(5), pp.7-23.
Park, D., & Shin, K. (2009). Saving, investment, and current account surplus in developing Asia. Asian Development Bank Economics Working Paper Series, (158).
Pebrianto, F. (2017). Indonesia Experiences Early Deindustrialization: Indef. [online] Tempo. Available at: https://en.tempo.co/read/844902/indonesia-experiences-early-deindustrialization-indef [Accessed 4 Nov. 2019].
Rodrik, D. (2015). Premature deindustrialization. Journal of Economic Growth, [online] 21(1), pp.1-33. Available at: https://www.nber.org/papers/w20935 [Accessed 4 Nov. 2019].
Rowthorn, R. and Coutts, K. (2013). De-industrialisation and the balance of payments in advanced economies. Centre for Business Research, University of Cambridge.
Tregenna, F. (2015). Deindustrialisation, structural change and sustainable economic growth, MERIT Working Papers 032, United Nations University - Maastricht Economic and Social Research Institute on Innovation and Technology (MERIT).
Wong, E. and Foo, K. (2018). Reversing premature deindustrialisation. [online] The Edge Markets. Available at: https://www.theedgemarkets.com/article/reversing-premature-deindustrialisation [Accessed 4 Nov. 2019].
เศรษฐพุฒิ สุทธิวาทนฤพุฒิ กฤษฏิ์ ศรีปราชญ์ จารีย์ ปิ่นทอง
ศิริกัญญา ตันสกุล รุจา อดิศรกาญจน์ แพรวไพลิน วงษ์สินธุวิเศษ, และ พิมพ์อร วัชรประภาพงศ์. (2017). การลงทุนเอกชนของไทย (2) : ทำไมระดับการลงทุนภาคเอกชนไทยต่ำ ควรทำและไม่ควรทำอะไร | ThaiPublica. [online] ThaiPublica. Available at: https://thaipublica.org/2017/08/8-facts-about-thai-investment2/ [Accessed 4 Nov. 2019].
เสาวณี จันทะพงษ์ นครินทร์ อมเรศ สมบูรณ์ หวังวณิชพันธุ์ ธนันธร มหาพรประจักษ์ และปาณิศาร์ เจษฎาอรรถพล. (2015). กระบวนการปรับโครงสร้างเศรษฐกิจไทยในปัจจุบันและทิศทางข้างหน้า: วิเคราะห์จากมุมมองตลาดแรงงาน. สายนโยบายการเงิน ธนาคารแห่งประเทศไทยAvailable at: https://www.bot.or.th/Thai/MonetaryPolicy/EconomicConditions/AAA/Presentation_Labour.pdf [Accessed 4 Nov. 2019].
I would like to express my appreciation to Sukjai Wongwaisiriwat, whose guidance and insights have made great contributions to this article. I would also like to thank Watsaya Limthammahisorn, Puntharik Supaarmorakul, Atipong Saikaew, members of Macroeconomics and Forecasting Division, and the FAQ editorial team for their helpful comments.
Kunlanan Chuntongvirat, Economist, Economic and Policy Department, Monetary Policy Group
Table 2 Summary of empirical studies