Green certification, e-commerce, and low-carbon economy for international tourist hotels

Abstract

Increasing population and over-consumption are placing unprecedented demands on agriculture and natural resources. The Earth is suffering from global warning and environmental destruction while our agricultural systems are concurrently degrading land, water, biodiversity, and climate on a global scale. For a sustainable future, green certification, e-commerce, and environment education can boost low-carbon economy with decreasing carbon emissions, but very few researches address them for the hotel industry. This research studies the performance impact of e-commerce, international hotel chain, local hotel chain, and green certification for carbon emission reductions of international tourist hotels of Taiwan. It reveals that, after a sufficiently long time, there is an improvement in the environmental and economic performance of the green-certified hotel group. In addition, it reveals that, as recommended by the operation policy, the international hotel chain group together with e-commerce has better performance than local hotel chain. It is also discussed how to sustain the continuing improvement in low-carbon performance of the hotel industry.

Appendices

Appendix 1

Table 4 Correlations between output and potential input variables by SPSS linear regression

Table 5 Coefficients between output and potential input variables by SPSS linear regression

Appendix 2

Table 6 Hotel chain/consortia systems, members in Taiwan, and requirements for memberships, green hotel policies

Appendix 3

Table 7 Green certifications and respective conditions to pass certification

Appendix 4. Methodology

To model the technology to produce desired outputs with jointly undesired outputs, some conditions are needed. If we denote desired outputs by $y\in {R}_{+}^M$, undesired outputs by $b\in {R}_{+}^I$, and inputs by $x\in {R}_{+}^N$, we can denote the output sets as:

$$ D(x)=\left\{\left(y,b\right):x\ \mathrm{can}\ \mathrm{produce}\ \left(y,b\right)\right\}. $$

(1.1)

The reduction of undesired outputs is costly and can be modeled as:

$$ \left(y,b\right)\ni D(x)\ \mathrm{and}\ 0\le \theta \le 1\ \mathrm{imply}\left(\theta y,\theta b\right)\ni D(x). $$

(1.2)

The desired outputs jointly produced with the undesired outputs can be modeled by:

$$ \mathrm{if}\ \left(y,b\right)\ni D(x)\ \mathrm{and}\ b=0\ \mathrm{then}\ y=0. $$

(1.3)

Based on conditions 1.1~1.3, Chung et al.’s definition for the ML index of productivity between period t and t + 1 is as:

$$ {\displaystyle \begin{array}{l}{\mathrm{ML}}_t^{t+1}=\frac{{\left[\left(1+D{\prime}_0^{t+1}\left({x}^t,{y}^t,{b}^t;{y}^t,-{b}^t\right)\right)\left(1+D{\prime}_0^t\left({x}^t,{y}^t,{b}^t;{y}^t,-{b}^t\right)\right)\right]}^{1/2}}{{\left(1+D{\prime}_0^{t+1}\left({x}^{t+1},{y}^{t+1},{b}^{t+1};{y}^{t+1},-{b}^{t+1}\right)\right)\left(1+D{\prime}_0^{t+1}\left({x}^{t+1},{y}^{t+1},{b}^{t+1};{y}^t,-{b}^{t+1}\right)\right)}^{1/2}}\\ {}=\frac{{\left[\left(1+D{\prime}_0^t\left({x}^t,{y}^t,{b}^t;{y}^t,-{b}^t\right)\right)\right]}^{\ast }}{\left(1+D{\prime}_0^{t+1}\left({x}^{t+1},{y}^{t+1},{b}^{t+1};{y}^{t+1},-{b}^{t+1}\right)\right)}\\ {}\begin{array}{l}\frac{{\left[\left(1+D{\prime}_0^{t+1}\left({x}^t,{y}^t,{b}^t;{y}^t,-{b}^t\right)\right)\left(1+D{\prime}_0^{t+1}\left({x}^{t+1},{y}^{t+1},{b}^{t+1};{y}^{t+1},-{b}^{t+1}\right)\right)\right]}^{1/2}}{{\left(1+D{\prime}_0^t\left({x}^t,{y}^t,{b}^t;{y}^t,-{b}^t\right)\right)\left(1+D{\prime}_0^t\left({x}^{t+1},{y}^{t+1},{b}^{t+1};{y}^{t+1},-{b}^{t+1}\right)\right)}^{1/2}}\\ {}={\mathrm{ML}\mathrm{EFFCH}}_t^{t+1}\ast {\mathrm{ML}\mathrm{TECH}}_t^{t+1}\end{array}\end{array}} $$

where D′₀ is a directional distance function and

$$ {D}_0^{\prime}\left(x,y,b;g\right)=\sup \left\{\beta :\left(y,b\right)+\beta g\in D(x)\right\} $$

where g is a vector of “directions” in which outputs are scaled.

ML index uses directional distance function than traditional Malmquist index using Shephard’s output distance function. The distance functions can be illustrated by Fig. 1. The meaning of point A in efficiency frontier in Malmquist index can be modeled as to scale OC/OA by OA/OC which increases both desired and undesired outputs proportionally from point C. The meaning of point B can be modeled as to scale OC by CB in efficiency frontier in ML index by increasing desired outputs and decreasing undesired outputs in direction Og (the same as CB). From the difference, it means that to increase productivity growth in Malmquist index one may have an increasing cost of pollution as a by-product, which is not what we expected in a sustainability effort. Thus, ML index is preferred since it can be expected to increase productivity growth without increasing undesired outputs.

For the efficiency of calculation, ML index can be indirectly derived from Malmquist index. To associate ML index with Malmquist index can be achieved through their respective based Shephard’s output distance function and directional distance function. The relation between directional distance function and Shephard’s output distance function is as:

$$ {\displaystyle \begin{array}{l}{D}_0^{\prime}\left(x,y,b;y,b\right)=\sup \left\{\beta :{D}_0\left(x,\left(y,b\right)+\beta \left(y,b\right)\right)\leqq 1\right\}\\ {}=\left(1/{D}_0\left(x,y,b\right)\right)-1\end{array}} $$

where D₀(x, y, b) is Shephard’s output distance function used in Malmquist index.

Fig. 1

Distance functions (source: Chen 2013a, b)