Asia-Pacific Journal of Regional Science

, Volume 2, Issue 1, pp 195–209 | Cite as

Climate change impact on labor productivity in Thai manufacture

  • Kitti Limskul


Average labor productivity in the Thai manufacturing sector is hypothesized to have adverse effect from rising temperature. The study applied Business and Manufacturing Census 2012 and found that 10 out of 12 sub-sectors are statistically significant with an expected negative sign. If Thailand would face with climate change volatility, it is projected that the climate change raises temperature from Baseline scenario by 2.5, 5.5, 6.0, and 7.5%, respectively. Present value of the damage of climate change is estimated to be 95,519 million baht in 2020. The loss has increased to 160,335 million baht in 2050 (in current prices). Damage is expected to be 4.63% in 2020 and 3.95% in 2050 of gross output of manufacture (in 2012 prices), respectively. The government may use fiscal and monetary policy to reshape the cost–benefit of investing in the adaptation of roofing and internal air temperature control. The firm is recommended to implement a medical rehabilitation for affected employees at the firm level to avoid relocation and absenteeism of workers.


Climate change impact Temperature rises Labor productivity Output loss 

JEL Classification


1 Significance of study

Climate variation has induced rising trend of average temperature as a global phenomenon. Heat stress has indirectly induced emotional suppression in workplace. Heat stroke may have also rendered directly to fatigue and losing ability to work in a factory as well. The climate change impact on labor productivity is global phenomena on both developing countries and developed countries alike.

The objective of this study is first to estimate the empirical relationship of climate variation or temperature rising and industrial labor productivity in Thailand. Second, we would like to construct a model to forecast future cost of climate change in Thailand 2010–2050.

2 Reviews of literatures

The study by the Asian Development Bank (ADB) (2009)1 has reported that Southeast Asia would face severe climate change, especially in the lower Mekong river basin. Climate change would have impacted on the food supply security in the region. Moreover, climate change would have raised the sea level, the frequency of flood and drought. Koonthanakoonwong et al. (2010) have applied the MRI-GCM model.2 The study has forecasted that the degree of temperature would be increased by 0.85 and 2.76 °C, respectively, on average during 2015–2039 and 2075–2099, respectively. The following survey of the literature has hypothesized that the climate change may, directly and indirectly, affect health and labor productivity in the region as well.3

A study by Koehn and Brown (1985) has applied econometric estimation and concluded that temperature rising affected labor productivity. Park (2016) has confirmed that rising temperature above 90 F has negatively induced per capita payrol declining in the US. In addition, Seppänen et al. (2006) have applied a quadratic econometric model to measure climate change impact on construction sector in the US. The study has found that extreme temperature rise would have caused productivity decline in the sector significantly. Hübler et al. (2008) have enquired into the climate change impact on labor productivity and health of employees in Germany. The author has applied the Regional Climate Model (REMO) projecting climate scenarios 2050–2100 based on the IPCC’s assumption and found that rising temperature has caused declining of labor productivity and GDP. It is hypothesised also that temperature has negative relationship with labor productivity in manufactureing sector.

In addition, Park (2015) has hypothesized that climate change with heat stress and stroke may have an impulse response by human behavior. Labor may choose to shift time off work and relocate his workplace to avoid productivity loss. Thus, cause and effect may be uneasy to determine without proper decomposition analysis. The productivity decline constituted by both ‘quantity and quality’ aspects. Regarding quantity, it means that a declining in labor productivity is attributed by ‘shorten labor hour in workplace.’ It is a result of heat stress and heat stroke, respectively. Regarding quality of work achievement, rising temperature has induced fatigue and stress. Park (2016) assesses a long-run climate adaptation of regions within the United States to daily heat shocks. The study applies a panel of county-level payroll and weather data (1986–2012) to estimate the causal relationship between hot days and local labor productivity. The result establishes a lower bound on labor productivity-related climate impacts at the global level.

Kjellstrom et al. (2009) have estimated the direct impact of climate change on regional labor productivity. Global climate change will increase outdoor and indoor heat loads and may impair health and labor productivity for millions of working people. This study applies physiological evidence about effects of climate change on working populations. The study estimates the impact of climate scenarios on future labor productivity. In most regions, climate change will decrease labor productivity, under the simple assumption of no specific adaptation. Climate change will significantly impact on labor productivity, inducing employers invest in adaptive measures.

Somanathann et al. (2014), the study used plant-level panel 21,509 samples data during 1998–1999 and 2007–2008 from ASI (Annual Survey of Industry, by the Government of India) to match temperature with the location of plants of the IMD (Indian Meteorological Department) database. With additional data from weather stations on heat and humidity, the model computed daily WBGT (Wet Bulb Globe Temperature). The panel model estimates the effect of temperature increase on labor productivity and output. In addition, when the temperature exceeded 25 °C, the industrial production has declined with the frequency of absenteeism among workers. Labor-intensive manufacturer would have lost its industrial product of 3% as result of temperature increase by 1 °C given capital, labor, and rain amount. Failing to account for reduced labor productivity may underestimate the costs of climate change. In addition, firms may have to control climate effect in the workplace and give attention to worker attendance as result of heat stress.

Li et al. (2016) have applied surveyed data on construction sector hiring blacksmith labor in China PRC and given age of labor, year of experience, and temperature at the site and concluded that climate change has a significant impact on labor productivity.

Langkulsen et al. (2010) have selected to survey labor productivity of industrial estates in Thailand located in the vicinity of Bangkok Metropolitan. The representative survey samples are from manufacturing industry and construction, respectively. The study has found that rising temperature in the workplace has caused productivity declined by 10–60%.

Verisk (2015) has, however, constructed a model to estimate the relationship between rising temperature (number of hot days in a year up to 2040) among the Southeast Asian countries and productivity of labor. The model has predicted that the more humid day in a year would have caused a declining in labor productivity of Southeast Asian nations by 5–25% from the baseline. Furthermore, Thailand would have lost labor productivity of 12% as compared with baseline forecast.

In our study, we share our view with the studies mentioned above in both developing and developed countries. Thailand is part of the global phenomenon of climate change. Thus, we may not turn our blind eyes on the economic impacts. The study in Thailand mentioned above has not explored into manufacture-wide sub-sector. In our study, we would like to apply Industrial Survey data 2012 of the National Statistical Office exploring the relationship of temperature and labor productivity.

3 Methodology

In this study, the ‘labor productivity’ is defined as the ratio of gross output (Y) over employed labor (L) in each manufacturing firm or (Y/L). In logarithmic form, the change in average labor productivity can be shown by log(Y/L), where Y is gross output and L is employed labor, respectively. Thus, the change in labor productivity as a result of climate variation or ‘temperature change’ is defined as differentiation of δlog(Y/L)/δtemp = Δ(Y/L)/(Y/L) when temperature changes 1 °C.

The Bank of Thailand has reported the productivity trend in Thailand during 2001–2013 as follows: sector with rising labor productivity comprised manufacturing, retail and wholesale trade, hotel and restaurant, transportation, banking and finance, and education sector, respectively. The industry that has stagnant of labor productivity was agriculture, construction, health services, and real estates, respectively. The sector with declining average labor productivity was mining and quarrying and public sector, respectively. It is clear that rising and dropping of labor productivity are crucial to GDP growth of Thailand in the past decades. The labor productivity may also be hypothesized related to extreme temperature change in Thailand as well. Over the reported period 2001–2013, we hypothesize that labor productivity movement in Thailand may be correlated with temperature as well. We have plotted the simple relationship between them as shown below and found that labor productivity may have an adverse correlation with rising temperature (see Fig. 1).
Fig. 1

Hypothesis on the relationship between temperature and labor productivity index of manufacturing sector in Thailand 2001–2013

(Source: Plot form data on labor productivity index of the Bank of Thailand 2001–2013, and Temperature is from Meteorological Department, and Thailand Greenhouse Gas Management Organization (TGO).

3.1 Model

In our study, we have followed an investigation by Somanathan et al. (2014) who used Annual Survey of Industry by the Government of India. In our study, we also apply manufacturing census data 2012 by the National Statistical Office to estimate the effect of temperature increase on labor productivity and output. We will also construct a forecasting model to evaluate the gross output loss and productivity loss, respectively 2020–2050.

Here, production of output Y with a Cobb–Douglas technology is defined as
$$Y = AL\left( {T_{\text{l}} , L_{\text{o}} } \right)^{\alpha } E^{\beta } K^{\gamma }$$
where L, E, and K are labor, energy, and capital input in the production function and A is an exogenous shift parameter of technological change. In our study, we add another exogenous information of temperature T into composite labor input \(L\left( {T_{\text{l}} , L_{\text{o}} } \right)\), where \(L_{\text{o}}\) is labor input and Tl temperature at work place (Wet Bulb Globe Temperature: WBGT)4
The possibility that temperature would affect labor productivity is
$$L\left( {T_{\text{l}} , L_{\text{o}} } \right)^{ } = \left\{ {\begin{array}{l} {L_{{{\text{o}} }} \; \text{if} \;T_{\text{l}} \;{\text{is}}\;{\text{less}}\;{\text{than}}\;T_{\text{C}} } \\ {L_{\text{o}} {\text{e}}^{{ - \theta T_{\text{l}} }} \;{\text{if}}\;T_{\text{l}} \;{\text{is}}\;{\text{greater}}\;{\text{than}}\;T_{\text{C}} } \\ \end{array} } \right.$$
where TC is “Comfortable or Ambient Temperature” to work efficiently.

In Eq. (2), if measured temperature is less than TC, labor productivity would not be affected. If measured temperature exceeds TC, it is likely that labor productivity would decline exponentially. In other words, the relationship between climate variation and labor productivity is non-linear.

If the outdoor temperature is high owing to climate variations, the indoor temperature may be not comfortable for working condition. In Thailand, the temperature of 25–26 °C is regarded as comfortable for most of the Thais. The hottest month is in April, where average temperature is approximately 30–34°. Therefore, ‘Tempmax-26’ is temperature exceed the comfortable warmth of the workplace.

Equations (1) and (2) imply that even though the human capital investment via education and training as well as technological advances, and thereby capital intensity may have positive relationship with labor productivity growth. On the contrary, climate change represented by temperature exceeds comfortable level or ‘Tempmax-26’ is hypothesized to have an adverse effect on labor productivity.

The non-linear relationship which can explain the above statement is, therefore
$$B_{j}^{\text{LD}} = A_{j}^{ } *\left( {K/L} \right)_{j}^{{\theta_{j1} }} {\text{e}}^{{ - \theta_{j2} {\text{Temp}}_{\text{max} }\text{-}26_{ } }}$$
where \(B_{j}^{\text{LD}}\) is labor productivity measured by average product per employee (baht/person); \(A_{j}^{ }\) is technological shift parameter; \(\left( {K/L} \right)\) capital intensity (baht/person) assumed to be constant in short run; and ‘Tempmax-26’ is maximum temperature exceeding ambient level of temperature measured in degree Celsius.5

The model defined “Tempmax-26” as a weighted average of daily temperature. Since our data samples distributed across regional location, we have matched the weighted average of maximum temperature reported by the Meteorological Department, Thai government with our data samples by location specific of firms.

The econometric model in double logarithmic form is as follows:
$$\text{log} \left( {B_{j,t}^{\text{LD}} } \right) = \log A_{j} + \theta_{j1} \log (K/L)_{j,t}^{ } + \theta_{j2} \left( {{\text{Temp}}_{\text{max} }\text{-}26} \right) + \xi .$$

The expected sign of coefficients are: θj1 > 0; θj2 < 0; ξ ~ N(0,1).

3.2 Data

In our study, we have applied sample surveyed data from the “Business and Industrial Census 2012”, the National Statistical Office. The establishments of manufacturing sub-sector with at least 11 employees6 are selected in this study as follows: (1) foods, drinks, and tobacco; (2) garment and textile products; (3) paper and paper products; (4) petroleum and coke; (5) chemical and chemical products; (6) rubber and plastic products; (7) non-metallic products; (8) basic metal and steel products; (9) fabricated metal products; (10) transport equipment; (11) electrical equipment and machinery; and (12) furniture products, respectively (Table 1).
Table 1

Description of selected data used in study.

Source: Industrial Census 2012, National Statistical Office, Thailand


Gross output (million baht per year)

Total labor (persons)

Net fixed asset (million baht)

Average productivity (million baht/person)

Asset-labor ratio (million baht per person)







Food drinks and tobacco






Textiles and garment






Paper and paper product






Petroleum and coke






Chemical and product






Rubber and plastic products






Non-metallic products






Basic metal, steel products






Fabricated metal products






Transportation equipments






Electrical machinery












In 2012, the manufacturing industry gross output was 7.08 trillion baht. It employed 3.7 million persons with net fixed asset of 3.8 trillion baht. The average productivity per person was 1.91 million baht and the asset per labor ratio was 1.03 million baht per person, respectively. The labor productivity was high in the petroleum and coke, basic metal and steel products, and transportation equipments sub-sector, respectively. The labor productivity was quite low in the Textiles and Garment, and the Furniture sub-sector.

4 Empirical result

4.1 Estimated coefficients of impact of climate change on labor productivity

The table below shows econometric estimations on impact of the climate change. It is relationship of rising temperature above or below the ambient temperature in the workplace. The negative sign implies that average labor productivity was affected by increasing heat exceeding comfortable level in workplace. The positive sign means that the labor productivity increase as a result of suitable temperature is dominant. The positive sign of ‘rain day’ coefficient means productivity increase owing to coolness of rainy day. Most of the manufacture sub-sector is affected by the climate change, where temperature exceeds the comfortable level. The ‘textile and garment’ and ‘petroleum and coke’ sectors have ‘positive’ signs in their climate coefficient and their rain day coefficient too. Their labor productivity is positively related to the comfort level which is lower than the maximal warmth and coolness of rainy days. Perhaps, the pleasant rainy days had brought about lower maximal temperature in their locations.

The econometric estimation with simple linear regression model is robust with significant levels. The climate coefficient and ‘rain day’ are statistically different from zero. As data are cross section in nature, we have obtained low R2 as expected. This means if some variables are added to the model, explanatory power may increase. Climate variation is a stochastic process thus with pooling of time series to this cross-sectional data random and fixed effect may give rise to better explanatory power. Burke et al. (2011, p. 11) have pointed out that the estimates that ignore climate uncertainty appear particularly likely to understate the climate impact. The paper has proposed to include the stochastic variable via a covariance into the regression equation. The future research can be revised further following this suggestion.

The estimated coefficients of the average labor productivity in the overall manufacturing sector with respected to the climate change variable are significant with negative sign (− 0.022). In addition, the rain day variable has positive relationship (0.019) with average labor productivity. All sub-sectors had negative relationship between climate variation and average productivity of labor, except the petroleum and coke, and textile and garments which had perverse sign and not significant. All sub-sectors except the basic metal and steel product have positive relationship with rain day variable.

The basic metal and steel product was adversely affected by climate variations as shown by estimated coefficient (− 0.366). It had, however, a negative relationship with rain day (− 0.012) but with much smaller the heat impact. This sub-sector has average labor productivity of 4.73 million baht per person. Thus, climate change may have significant implications on sectors which utilize intense energy and heat content compared with other sub-sectors. This is postulated also in the case of transportation equipment and the electrical machinery for instance (Table 2).
Table 2

Estimated coefficients of climate change impact on Manufacture Industry in Thailand 2012

Dependent variable

Explanatory variables


Labor productivity log (Y/L)


Log (K/L)


Rain day

R 2



3.974*** (0.246)

0.482*** (0.005)

− 0.022** (0.011)

0.019*** (0.001)



Food drinks and tobacco

7.073*** (0.448)

0.479*** (0.008)

− 0.084*** (0.020)

0.002 (0.001)



Textiles and garment

1.374* (0.819)

0.357*** (0.016)

0.020 (0.037)

0.039*** (0.003)



Paper and paper product

3.166*** (1.104)

0.334*** (0.034)

− 0.152*** (0.038)

0.062*** (0.008)



Chemical and product

− 0.810 (1.234)

0.63*** (0.031)

− 0.111*** (0.044)

0.059*** (0.009)



Petroleum and coke

− 2.686 (3.383)

0.724*** (0.076)

0.166 (0.142)

0.038*** (0.013)



Rubber and plastic products

6.887*** (0.860)

0.345*** (0.019)

− 0.066*** (0.038)

0.022*** (0.003)



Non-metallic products

6.608*** (0.600)

0.480*** (0.012)

− 0.067** (0.027)

0.006*** (0.002)



Basic metal, steel products

13.217** (1.567)

0.504*** (0.034)

− 0.366*** (0.066)

− 0.012* (0.007)



Transportation equipment

10.323*** (0.731)

0.328*** (0.016)

− 0.114*** (0.031)

0.001 (0.003)



Fabricated metal product

9.65*** (0.589)

0.260*** (0.014)

− 0.097*** (0.025)

0.006*** (0.002)



Electrical machinery

1.725 (1.390)

0.430*** (0.043)

− 0.180*** (0.053)

0.067*** (0.009)




4.897*** (1.153)

0.284*** (0.026)

− 0.110** (0.050)

0.018*** (0.004)



Estimated from Industrial Census 2012. Standard error shown in parenthesis

Y gross output, Y/L average labor productivity, K/L fixed asset per labor, T-26 Tempmax-26, N no. of samples

*** Significant at 99% ** 95% * 90%

4.2 Valuation of partial impact of climate change on labor productivity

The overall manufacturing average labor productivity was negatively affected by rising temperature above ambient level surrounding the workplace from its past trend. As a semi-logarithmic function, temperature increases 1 °C, and the average labor productivity would decline by − 2.2% per year (− 0.022 × 100) on average. It is consistent with a study by Somanathann et al. (2014) when the temperature exceeded 25 °C the industrial output has declined with the frequency of absenteeism among workers. The study has found that labor-intensive manufacture would have lost its industrial product of 3% as result of temperature increase by 1 °C given capital, labor, and rain amount.

Regarding estimated coefficients’ magnitude, as maximum temperature increases 1 °C, it is found that average labor product in manufacture sub-sectors is affected as follows:
  1. 1.

    Sub-sectors with high climate change impact (rising temperature in the workplace) on labor productivity comprise basic metal and steel products (− 36.6%), electrical machinery (− 18.0%), paper, and paper product (− 15.2%), respectively.

  2. 2.

    Sub-sectors with moderate impact comprise chemical and products (− 11.1%), transport equipments (− 11.4%), and furniture (− 11.0%), foods, drinks, and tobacco products (− 8.4%), non-metallic products (− 6.7%), rubber and products (− 6.6%), and fabricated metal products (− 9.7%), respectively.

As the temperature of the working place rises 1 °C from the base level in Thailand, the ‘average labor product’ of Thai manufacture sector in 2012 would decline − 41,674 baht per person. The loss in sub-sector ‘basic metal and steel product’ is − 1,731,085 baht per person, while the minimum loss in sub-sector Furniture product is − 73,582 baht per person, respectively (Table 3).
Table 3

Loss of average labor productivity in Thai manufacturing as result of rising temperature 1 °C over regional location of firms in 2012



Loss of average labor productivity per 1 °C (baht/person)



− 42,046


Basic metal, steel products

− 1,731,085


Transportation equipment

− 484,256


Electrical machinery

− 483,435


Chemical and product

− 315,415


Paper and paper product

− 301,267


Rubber and plastic products

− 174,299


Food drinks and tobacco

− 142,831


Fabricated metal product

− 124,311


Non-metallic products

− 107,828



− 73,582

We omit sub-sector petroleum and coke sample, and textile and garment owing to perverse signs and not significant

The estimated loss is regarding baht per 1 °C. It is derivative of \(\delta {{\log \left( {B_{j,t}^{\text{LD}} } \right)}/{\delta ({\text{Temp}}_{ \text{max} } {\text{-}26})}} = - \theta_{j 2}\)

Only selected sub-sectors

Climate change which is a global phenomenon has induced damage on gross output as well. Loss of labor productivity has adversely influenced the output loss of overall manufacture sector. In a static analysis above, the ‘rain day’ has contributed positively to counter the negative impact of temperature rise beyond comfortable level in the workplace. The coefficients estimate would be applied to forecast the output loss over time.

Since our detail manufacturing data is cross section in nature, it may not represent future trend adequately as time series of manufacturing census is not available in our study. We, therefore, collect a time series of aggregate manufacturer data from the National Account statistics. Hsiao (2003, p. 287) pointed out an application of ‘Likelihood-ratio test for the log joint likelihood function of a single cross section consisting of N units and a time series extending over T time periods to decide whether sample data are pooled. In our study, we estimate a linear regression model of climate change impact on overall manufacturing and measure the estimated coefficients of implications.

We have applied the historical net capital stock at 1988 price (Kt) and labor (Lt), and temperature (2001–2013) to determine the average labor productivity. In logarithmic form, we obtain the estimated coefficients of the equation:
$$\begin{array}{*{20}c} {\mathop {{\text{Log}}\left( {{\text{GDP}}/L} \right) = 0.63}\limits_{\begin{subarray}{l} \left( {t{\text{-val}} \,=\, 1.0 1;} \right) \\ \left( {{\text{std}}.{\text{err}}\, =\, 0. 6 2} \right) \end{subarray} } } {\mathop { \, + \, 1.04 \, *{\text{Log}}\left( {K/L} \right)}\limits_{\begin{subarray}{l} \left( {t{\text{-val }} \,=\, 10. 3 9} \right) \\ \left( {{\text{std}}.{\text{err}} \,=\, 0. 10} \right) \end{subarray} } } {\mathop { \;- \;0.018*\left( {{\text{Temp}}_{ \text{max} } { - }26} \right)}\limits_{\begin{subarray}{l} \left( {t{\text{-val }} = \, - \; 2. 2 6} \right) \\ \left( {{\text{std}}.{\text{err}} \,=\, 0.00 8} \right) \end{subarray} } } \\ \end{array} .$$

Data source: National Accounts statistics. R2 = 0.915; D.W. = 1.665; S.E of Regression = 0.048; and Period of estimation 2001–2013.

The coefficient of ‘Tempmax-26’ is ‘− 0.018’ in the time-series model is close to the coefficient of cross-sectional estimate of ‘− 0.022’, as shown in Table 2.7 It may imply that cross-sectional and time-series data are consistent and robust. As a result, this confirms a negative impact of rising temperature on the average labor productivity in short run and over time. The pooling of time-series and cross-sectional data may be a future research.8

4.3 Forecasting result of climate change impact on labor productivity and output loss 2020–2050

We apply the estimated relationship of climate variation and labor productivity at a point in time (2012) from model estimation above. Now, we assume that temperature in Thailand would rise as result of climate change during 2020–2050 towards the threshold of ‘an increment of 2 °C from past trend’ as conventionally understood in most of climate change studies.9

The base period temperature, the maximum temperature, the rain day, and rain amount series are predicted for the period of 2020–2050. The mean value of the base temperature is 38.83 °C; the maximum temperature is 40.78 °C with an increase on average 1.95 °C from past trend. The rain amount of 1695.8 mm and rain day 128.5 days (measured at mean) are assumed in the forecasting model (Table 4).
Table 4

Scenario of climate change represented by related variables, Thailand 2020–2050


Average temperature rise from past trend (°C)

Base temperature (°C)

Maximum temperature (°C)

Rain amount (mm)

Rain day













Std. dev.







− 0.45

− 0.52

− 0.73

− 0.85

− 0.77







Incremental temperature was advised by Thailand Greenhouse Gas Management Organization, and study by Koonthanakoonwong et al. (2010). The temperature level is extrapolated from actual observable temperature report by Department of Meteorology, Thailand

The equation of overall manufacturing sector is assumed to represent the sum of the sub-sectors in the forecasting model. The model has applied the predicted exogenous variables mentioned earlier. The result is reported in year 2020–2050 in Table 5. The cost of climate change or damage is measured as the difference between scenario 1 and baseline projection.
Table 5

Evaluation of overall cost of climate change impact on manufacturing industry in Thailand 2020–2050: applying cross-sectional data as based with model forecasting


Gross output (million baht)

Gross damage, (million baht)

(3) = (2) − (1)

Present value, of damage (million baht)

(4) = (3)/(1 + r)t

Damage per temperature change of 1 °C (million baht)


Damage cost in million baht per gross output

(6) = %(4)/(1)

Baseline (1)

Scenario 1 (2)




− 116,214

− 95,519

− 46,486

− 4.63%




− 287,144

− 159,441

− 52,208

− 6.44%




− 405,305

− 152,037

− 67,551

− 4.75%




− 632,697

− 160,335

− 84,360

− 3.95%

The climate change scenario induces a temperature increase from past trend not exceeding 2 °C at long-term mean value

In our study, we utilize detail cross-sectional sample data of the Industrial Census 2012, the National Statistical Office. We did not have detail sample data of 1989, 1999, 2000, and 2007. The NSO has published only the tabulated data tables. It will be future effort to try the full sample estimation

The ‘Present value of Damage’ at year t0 is defined merely as = \(({\text{Damage}}\; {\text{value}}\;{\text{at}}\;{\text{year}}\;t)/\left( {1 + r} \right)^{t}\), where r is given to be 4% per year throughout forecasting horizon. We have measured the mean value of gross output of manufacture and gross output per head of manufacture labor which would be affected by the climate change, i.e., a damage under the “scenario 1” over the forecasting period of 2020–2050. Climate change has been forecasted to raise temperature from baseline scenario by 2.5, 5.5, 6.0, and 7.5%, respectively. The present value of the damage is estimated to be 95,519 million baht in 2020. The cost of climate change has increased to 160,335 million baht in 2050 (in current prices). Damage is estimated to be 4.63% in 2020 and 3.95% in 2050 of gross output of manufacture (in 2012 prices), respectively (Table 4).

The overall cost of climate change is a constituted damage of all sub-sectors (except the petroleum and coke which has perverse sign). The average labor productivity is assumed to follow the order of calculated magnitude, as shown in Table 3. For instance, the first five sub-sector damages are the basic metal and steel products, transportation equipment, electrical machinery, chemical and products, and paper and paper product, respectively.

5 Conclusions and recommendations

The climate change is a global phenomena, and it affects labor productivity worldwide including Thailand. In our study, we have applied data from Manufacturing Census 2012 to estimate the relationship between temperature change as proxy of climate variation and average labor productivity (measured in terms of gross output-labor ratio) of manufacturing sub-sectors. The temperature officially reports by the Department of Meteorological is matched with a regional location of surveyed firms. We have rejected the null hypothesis of climate impact on average labor productivity in ten sub-sectors. We alternatively have found a statistical significance negative relationship between rising temperature and average product of labor at the level of the manufacturing sub-sector. Ten sub-sectors are statistically significant with an expected negative sign. Most of these sub-sectors have positive relationship with ‘rain day’ at significant level.

The study has found that if temperature increases 1 °C, the average manufacture’s labor productivity will decline by − 42,046 baht per person. Damage by each sub-sector is different according to their labor’s resistant to stress and illness from heat stroke.

If Thailand would face with climate change volatility, it is projected that the climate change raises temperature from baseline scenario by 2.5, 5.5, 6.0, and 7.5%, respectively. The present value of the damage of climate change is estimated to be 95,519 million baht in 2020. The loss has increased to 160,335 million baht in 2050 (in the current prices). Damage is expected to be 4.63% in 2020 and 3.95% in 2050 of gross output of manufacture (in 2012 prices), respectively.

This partial equilibrium result has some shortcomings. The study applies cross-sectional data of 2012. Thus, the model may not capture total long-run effect of climate change impact. We have test the consistency of estimation results using the time-series data report in the National Accounts of Thailand. The estimated coefficient of time-series model is consistent with cross-sectional data estimate for the overall manufacturing industry. The sign of the expected Coefficient and magnitude is approximately the same in measuring the climate change impact. The model with the panel data pooling between time-series and cross-sectional data with proper matching between sub-sector and location specific of climate variation may raise explanatory power of the estimates to considerable extent. This may be further study when data are available across industry location in detail over specified years.

In our current study, we have added explanatory variable ‘rain day’ to represent the influence of wet and humid. This may tone down the heat effect of temperature rising in our estimation result and forecasting. It is the coolness of workplace approximating the ‘Wet Bulb Globe Temperature’ definition. We hypothesize that working place temperature that exceeds the comfortable temperature is differentiated among manufacturing firms’ location. Adding these explanatory variables into our model may tone down the size of estimated impact coefficients. Despite insufficient data on WBGT in our model, the cross-sectional and time-series estimates are consistent and may be regarded as the ceiling of the impact of climate change on the manufacturing sector in Thailand.

This study would propose the following recommendations on adaptation efforts: (1) private sector which runs manufacturing firms may need to invest in the structure of roofing and internal air temperature control to create the comfortable working environment. It may tone down the direct effect of heat wave from sun. The rising temperature owing to climate change may hardly hit Thailand in coming decades. The proper design of air ventilation flow through the system, by cooling water pond and shading outside factory dome can be an appropriate solution of climate change adaptation. Government’s initiative for a design contest and promote research on material for building structure would be helpful with incentives for adaptation effort. (2) Initiation of government on cost–benefit of adaptation with tax incentive and interest subsidy for the factory that introduced renewable energy systems such as solar panel, heat exchanger, and other material to build-in the system to reduce heat wave. (3) The firm is recommended to implement a medical rehabilitation for affected employees at the firm level to avoid relocation and absenteeism of workers.


  1. 1.
  2. 2.

    A general circulation model (GCM) is a type of climate model. It employs a mathematical model of the general circulation of a planetary atmosphere or ocean. The model belongs to the Meteorological Research Institute, (MRI), Meteorological Agency Japan. Koonthanakoonwong et al. (2010) had organized a joint research on climate change impact in Thailand with MRI by applying GCM. The prediction scenarios of temperature used in the model are referred in study organized by Limskul (2011). The predicated temperature is again used in this study.

  3. 3.
  4. 4.

    This is unit of temperature measurement at work place.

  5. 5.

    Official temperature data series is published by the Department of Meteorological comprises maximum temperature and minimum temperature.

  6. 6.

    This is not inclusive of construction, transportation, banking and finance and agriculture sector.

  7. 7.

    Average labor productivity in Table 2 based on cross-sectional data is measured in ‘Gross Output per labor’ while the time-series data model is measured in ‘Gross Domestic Product or GDP per labor’.

  8. 8.

    The National Statistical Office has Business and Manufacturing Industry Survey of 1989, 1999, 2000, 2007 and 2012. The NSO has published only the tabulated data tables. It will be future effort to try the full sample estimation.

  9. 9.

    It is noted by Stern Review (2007) that climate change could have very serious impacts on growth and development. If no action is taken to reduce emissions, the concentration of greenhouse gases in the atmosphere could reach double its pre-industrial level as early as 2035, virtually committing us to a global average temperature rise of over 2 °C.


  1. Asian Development Bank (2009) The economics of climate change in Southeast Asia: a regional review. Accessed 13 Sept 2017
  2. Burke M, Dykema J, Lobell D, Miguel E, Satyanath S (2011) Incorporating climate uncertainty into estimates of climate change impacts, with applications to the US and African agriculture. National Bureau of Economic ResearchGoogle Scholar
  3. Hsiao C (2003) Analysis of panel data, 2nd edn. Cambridge University Press, CambridgeCrossRefGoogle Scholar
  4. Hübler M, Klepper G, Peterson S (2008) Costs of climate change: the effects of rising temperatures on health and productivity in Germany. Ecol Econ 68(1–2):381–393. CrossRefGoogle Scholar
  5. Kjellstrom T, Kovats RS, Lloyd SJ, Holt T, Tol RS (2009) The direct impact of climate change on regional labor productivity. Arch Environ Occup Health 64(4):217–227CrossRefGoogle Scholar
  6. Koehn E, Brown G (1985) Climatic effects on construction. J Constr Eng Manag 111(2):1985CrossRefGoogle Scholar
  7. Koonthanakoonwong S et al. (2010) Collected MRI-GCM data for Thailand. Technical Report. Chulalongkorn University, ThailandGoogle Scholar
  8. Langkulsen U et al (2010) Health impact of climate change on occupational health and productivity in Thailand. Glob Health Action 3:5607. CrossRefGoogle Scholar
  9. Li X, Chow KH, Zhu Y, Lin Y (2016) Evaluating the impacts of high-temperature outdoor working environments on construction labor productivity in China: a case study of rebar workers. Build Environ 95:42–52CrossRefGoogle Scholar
  10. Limskul K (2011) A Master Plan of the Climate Change Impact in Thailand submitted to the National Economic and Social Development Board, Thai government. There 6 supporting reports as follows: (1) Main Climate Model, (2) Regional climate model (MRI_GCM) on precipitation forecast for Thailand, (3) Economic impact of the climate change applying I-O analysis, (4) Energy model, (5) Agriculture impact study (6) Climate impact and adaptation a field survey. The Master Plan was unpublished reports in ThaisGoogle Scholar
  11. Park J (2015) Labor productivity impacts from climate change: Implications for global poverty. World Bank Climate Change and Poverty Conference 2015Google Scholar
  12. Park J (2016) Will we adapt? Temperature shocks, labor productivity, and adaptation to climate change in the United States (1986–2012). Discussion paper 16–81, The Harvard project on climate agreements, Harvard UniversityGoogle Scholar
  13. Seppänen O, Fisk WJ, Lei Q (2006) Effect of temperature on task performance in an office environment. Ernest Orlando Lawrence Berkeley National LaboratoryGoogle Scholar
  14. Somanathan E et al (2014) The impact of temperature on productivity and labor supply: Evidence from Indian manufacturing. Discussion paper 14-10. Indian Statistical Institute. Delhi Economic and Planning Unit. School of Economics DelhiGoogle Scholar
  15. Stern N (2007) The economics of climate change: the Stern review. Cambridge University Press, Cambridge.
  16. Verisk M (2015) Heat stress poses a threat to SE Asia productivity. Accessed 1 May 2017

Copyright information

© The Japan Section of the Regional Science Association International 2018

Authors and Affiliations

  1. 1.Faculty of EconomicsSaitama UniversitySaitamaJapan

Personalised recommendations