• 沒有找到結果。

第六章 結論與未來研究方向

第一節 結論

立 政 治 大 學

N a tio na

l C h engchi U ni ve rs it y

84

第六章 結論與未來研究方向

第一節 結論

射線距離函數模型由於忽略了非意欲產出,傾向高估生產單位的效率分數。

方向距離函數模型估計結果顯示為技術進步,然而射線距離函數模型卻顯示為技 術退步,兩者結果截然不同。由於射線距離函數模型忽略了非意欲產出,且在多 數的資料下都不能滿足射線距離函數的先天性質,因此較無參考價值。由此可見,

射線距離函數在我們實證中並非理想的模型,較好的方式是以方向距離函數模型 進行估計。

本文實證結果中,方向距離函數模型中若未加入限制條件,將低估無效率值、

高估效率分數與技術進步率;未加入環境變數,將低估無效率值與效率分數、高 估帄均技術進步率。另一方面,加入限制條件後能同時滿足方向距離函數在各樣 本的計算下為非負,符合方向距離函數的先天設定。相較之下,未加入限制條件 則會在某幾筆資料下得出方向距離函數為負的極端情況。最後,計算各個所得類 別的帄均技術進步率估計值後,僅有同時加入限制條件與環境變數的方向距離函 數模型顯示為越低所得國家,其技術進步率估計值越高,較符合經濟方面的解釋。

因此對生產單位進行效率分析時,較適當是採用同時加入限制條件與環境變數的 方向距離函數模型。

‧ 國

立 政 治 大 學

N a tio na

l C h engchi U ni ve rs it y

85

第二節 未來研究方向

至於未來的研究方向,首先可以嘗詴使用連續型的環境變數取代本文所使用 的類別變數。由於本文的類別變數僅有分成四類,無法貼切地給定所有無效率項 的先驗分配參數。而且使用連續型環境變數的情況下,能透過其係數觀察無效率 項與環境變數之間的關聯。然而,連續型的環境變數將導致其條件分配難以推導 的問題,在使用 WinBUGS 以外的統計軟體時,將會需要使用更複雜的參數抽樣 方法。若是使用 WinBUGS 進行分析,仍然需要找到良好的起始值,而良好的起 始值在有限制條件的情況下將會更難以尋找。

再者,若混和所有國家資料估計單一前緣,隱含假設為所有國家都具有同樣 的生產技術。估計單一前緣的優點為可以在國家之間進行比較,因為所有國家都 是以該前緣為基準;而缺點為容易違反生產技術相同的假設。根據本文在方向距 離函數與射線距離函數模型的實證結果,都指出高所得國家具有較高的效率分數 估計值,意即其生產技術較高,而低所得國家其生產技術較低。因此較好的方式 應為在不同所得類別下,估計不同的前緣,優點在於能夠符合該類別的國家生產 技術相同的假設,而缺點為不同類別的國家無法互相比較,因為比較的基準不同。

估計各所得類別的前緣後,透過概似比檢定 (likelihood ratio test) 即可比較各前 緣之間的差異是否顯著,檢驗是否滿足各所得類別其生產技術相同的假設。

‧ 國

立 政 治 大 學

N a tio na

l C h engchi U ni ve rs it y

86

附錄

1. 方向距離函數模型的單調性限制推導

滿足單調性限制,方向距離函數需對於產出𝑦為非遞增、對於非意欲產出𝑏為 非遞減、對於投入𝒙為非遞減。因此,我們將方向距離函數(9)式對𝑥1, 𝑥2, 𝑦, 𝑏偏微 分

𝜕𝐷⃗⃗ 0(𝑦,𝑏,𝑥,𝑡;1,;1)

𝜕𝑦 = 𝛽1+ 𝛽6𝑦 + 𝛽11𝑥1+ 𝛽12𝑥2+ 𝛽13𝑏 + 𝛽14𝑡 ≤ 0 (A.1)

𝜕𝐷⃗⃗ 0(𝑦,𝑏,𝑥,𝑡;1,;1)

𝜕𝑏 = 𝛽4 + 𝛽9𝑏 + 𝛽13𝑦 + 𝛽16𝑥1+ 𝛽18𝑥2+ 𝛽20𝑡 ≥ 0 (A.2)

𝜕𝐷⃗⃗ 0(𝑦,𝑏,𝑥,𝑡;1,;1)

𝜕𝑥1 = 𝛽2+ 𝛽7𝑥1+ 𝛽11𝑦 + 𝛽15𝑥2+ 𝛽16𝑏 + 𝛽17𝑡 ≥ 0 (A.3)

𝜕𝐷⃗⃗ 0(𝑦,𝑏,𝑥,𝑡;1,;1)

𝜕𝑥2 = 𝛽3+ 𝛽8𝑥2+ 𝛽12𝑦 + 𝛽15𝑥1+ 𝛽18𝑏 + 𝛽19𝑡 ≥ 0 (A.4)

由於𝛽1、𝛽6、𝛽11、𝛽12、𝛽13、𝛽14未在轉換後的模型出現,因此我們代入轉換性 質(11)-(15)式,可得

𝜕𝐷⃗⃗ 0(𝑦,𝑏,𝑥,𝑡;1,;1)

𝜕𝑦 = (𝛽4− 1) + 𝛽9𝑦 + 𝛽16𝑥1+ 𝛽18𝑥2+ 𝛽9𝑏 + 𝛽20𝑡 ≤ 0 (A.5)

𝜕𝐷⃗⃗ 0(𝑦,𝑏,𝑥,𝑡;1,;1)

𝜕𝑏 = 𝛽4 + 𝛽9𝑏 + 𝛽9𝑦 + 𝛽16𝑥1 + 𝛽18𝑥2+ 𝛽20𝑡 ≥ 0 (A.6)

𝜕𝐷⃗⃗ 0(𝑦,𝑏,𝑥,𝑡;1,;1)

𝜕𝑥1 = 𝛽2+ 𝛽7𝑥1+ 𝛽16𝑦 + 𝛽15𝑥2+ 𝛽16𝑏 + 𝛽17𝑡 ≥ 0 (A.7)

𝜕𝐷⃗⃗ 0(𝑦,𝑏,𝑥,𝑡;1,;1)

𝜕𝑥2 = 𝛽3+ 𝛽8𝑥2+ 𝛽18𝑦 + 𝛽15𝑥1+ 𝛽18𝑏 + 𝛽19𝑡 ≥ 0 (A.8)

而所有偶數的主子式為非負,參考 Morey (1986) 與 Feng and Serletis (2014)。在 我們的模型中,意即

參考O’Donnell and Coelli (2005),D(y, x, t)對於𝒙具有準击性,充分條件為 D(y, x, t)對於𝒙的鑲邊海森矩陣其所有鑲邊主子式為負,而鑲邊主子式之定義可

‧ 國

立 政 治 大 學

N a tio na

l C h engchi U ni ve rs it y

92

8 Bolivia 30 Honduras 52 Mexico 74 Sweden

9 Brazil 31 Hong Kong 53 Morocco 75 Switzerland

10 Bulgaria 32 Hungary 54 Mozambique 76 Syria

11 Cambodia 33 Iceland 55 Netherlands 77 Tanzania

12 Cameroon 34 India 56 New Zealand 78 Thailand

13 Canada 35 Indonesia 57 Niger 79 Trinidad & Tobago

14 Chile 36 Iran 58 Norway 80 Tunisia

15 China 37 Iraq 59 Pakistan 81 Turkey

16 Colombia 38 Ireland 60 Panama 82 Uganda

17

Congo, Dem.

Rep.

39 Israel 61 Paraguay 83 United Kingdom

18 Costa Rica 40 Italy 62 Peru 84 United States

19 Cote d`Ivoire 41 Jamaica 63 Philippines 85 Uruguay

20 Cyprus 42 Japan 64 Poland 86 Vietnam

21 Denmark 43 Jordan 65 Portugal 87 Zambia

22

Dominican

Republic

44 Kenya 66 Qatar 88 Zimbabwe

‧ 國

立 政 治 大 學

N a tio na

l C h engchi U ni ve rs it y

93

參考文獻

Assaf, A.G., R. Matousek and E.G. Tsionas, (2013). Turkish bank efficiency:

Bayesian estimation with undesirable outputs, Journal of Banking and Finance, 37, 506-517.

Atkinson, S.E. and J.H. Dorfman, (2005). Bayesian measurement of productivity and efficiency in the presence of undesirable outputs: crediting electric utilities for reducing air pollution, Journal of Econometrics, 126, 445-468.

Battese, G.E. and T.J. Coelli, (1995). A model for technical inefficiency effects in a stochastic frontier production function for panel data, Empirical Economics, 20, 325-332.

Breheny, P., (Retrieved September 26 2016). MCMC methods: Gibbs and Metropolis, from: http://web.as.uky.edu/statistics/users/pbreheny/701/S13/notes/2-28.pdf

Bokusheva, R. and S.C. Kumbhakar, (2014). A distance function model with good and bad outputs, from:

http://ageconsearch.umn.edu/bitstream/182765/2/Bokusheva-Distance_function_mod el_with_good_and_bad_outputs-258_a.pdf

Boyd, G.A., G. Tolley and J. Pang, (2002). Plant level productivity, efficiency, and environmental performance of the container glass industry, Environmental and Resource Economics, 23, 29-43.

‧ 國

立 政 治 大 學

N a tio na

l C h engchi U ni ve rs it y

94

Broeck, J. van den, G. Koop, J. Osiewalski and M.F.J. Steel, (1994). Stochastic frontier models: a Bayesian perspective, Journal of Econometrics, 61, 273-303.

Chambers, R.G., (2002). Exact nonradial input, output, and productivity measurement, Economic Theory, 20, 751-765.

Chung, Y.H., R. Färe and S. Grosskopf, (1997). Productivity and undesirable outputs:

a directional distance function approach, Journal of Environmental Management, 51, 229-240.

Cowles, M.K. and B.P. Carlin, (Retrieved September 26 2016). Markov chain Monte Carlo convergence diagnostics: a comparative review, from:

http://www.public.iastate.edu/~alicia/stat544/rr94-008.pdf

Färe, R. and S. Grosskopf, (2003). New directions: efficiency and productivity, US:

Springer.

Färe, R., S. Grosskopf, D. Noh and W. Weber, (2005). Characteristics of a polluting technology, Journal of Econometrics, 126, 469-492.

Feng, G. and A. Serletis, (2010). A primal Divisia technical change index based on the output distance function, Journal of Econometrics, 159, 320-330.

Feng, G. and A. Serletis, (2014). Undesirable outputs and a primal Divisia productivity index based on the directional output distance function, Journal of Econometrics, 183, 135-146.

‧ 國

立 政 治 大 學

N a tio na

l C h engchi U ni ve rs it y

95

Feng, G. and X.H. Zhang, (2012). Productivity and efficiency at large and community banks in the US: a Bayesian true random effects stochastic distance frontier analysis, Journal of Banking and Finance, 36, 1883-1895.

Fernández, C., G.M. Koop and M. Steel, (2002). Multiple output production with undesirable outputs: an application to nitrogen surplus in agriculture, Journal of the American Statistical Association, 97, 432-442.

Fernández, C., J. Osiewalski and M.F.J. Steel, (1997). On the use of panel data in stochastic frontier models with improper priors, Journal of Econometrics, 79,169-193.

Flegal, J.M., (2008). Monte Carlo standard errors for Markov chain Monte Carlo, from: http://www.faculty.ucr.edu/~jflegal/Final_Thesis_twosided.pdf

Griffin, J.E. and M.F.J. Steel, (2007). Bayesian stochastic frontier analysis using WinBUGS, Journal of Productivity Analysis, 27, 163-176.

Griffin, J. E. and M.F.J. Steel, (2008). Flexible mixture modeling of stochastic frontiers, Journal of Productivity Analysis, 29, 33-50.

Griffiths, W.E., C.J. O’Donnell and A.T. Cruz, (2000). Imposing regularity conditions on a system of cost and factor share equations, The Australian Journal of Agricultural and Resource Economics, 44, 107-127.

Huang, C.J. and J.-T. Liu, (1994). Estimation of a non-neutral stochastic frontier production function, Journal of Productivity Analysis, 5, 171-180.

‧ 國

立 政 治 大 學

N a tio na

l C h engchi U ni ve rs it y

96

Huang, H.C., (2004). Estimation of technical inefficiencies with heterogeneous technologies, Journal of Productivity Analysis, 21, 277-296.

Huang, T.-H., (2005). A study on the productivities of IT capital and computer labor:

firm-level evidence from Taiwan’s banking industry, Journal of Productivity Analysis, 24, 241-257.

Kleit, A.N. and D. Terrell, (2001). Measuring potential efficiency gains from

deregulation electricity generation: a Bayesian approach, Review of Economics and Statistics, 83, 523-530.

Koop, G., J. Osiewalski and M.F.J. Steel, (1994b). Hospital efficiency analysis with individual effects: a Bayesian approach, Center for Economic Research discussion paper, 9447.

Koop, G., J. Osiewalski and M.F.J. Steel, (1997). Bayesian efficiency analysis through individual effects: hospital cost frontiers, Journal of Econometrics, 76, 77-105.

Kurkalova, L. A. and A. Carriquiry, (2003). Input- and output-oriented technical efficiency of Ukrainian collective farms, 1989-1992: Bayesian analysis of a stochastic production frontier model, Journal of Productivity Analysis, 20, 191-211.

Lam, P., (Retrieved September 26 2016). MCMC methods: Gibbs sampling and the Metropolis-Hastings algorithm, from:

http://pareto.uab.es/mcreel/IDEA2015/MCMC/mcmc.pdf

‧ 國

立 政 治 大 學

N a tio na

l C h engchi U ni ve rs it y

97

Lee, J.-D., J.-B. Park and T.-Y. Kim, (2002). Estimation of the shadow prices of pollutants with production/environment inefficiency taken into account: a nonparametric directional distance function approach, Journal of Environmental Management, 63, 365-375.

Lin, E.T.J. and L.W. Lan, (2010). Measuring firm-specific efficiencies with Bayesian stochastic distance function, 2010 International Conference on Asia Pacific Business Innovation and Technology Management.

Morey, E.R., (1986). An introduction to checking, testing, and imposing curvature properties: the true function and the estimated function, Canadian Journal of Economics, 19, 207-235.

O’Donnell, C.J. and T.J. Coelli, (2005). A Bayesian approach to imposing curvature on distance functions, Journal of Econometrics, 126, 493-523.

Orea, L., (2002). Parametric decomposition of a generalized Malmquist productivity index, Journal of Productivity Analysis, 18, 5-22.

Osiewalski, J. and M.F.J. Steel, (1998). Numerical tools for the Bayesian analysis of stochastic frontier models, Journal of Productivity Analysis, 10, 103-117.

Sinharay, S., (2003). Assessing convergence of the Marlov chain Monte Carlo algorithms: a review, from:

http://www.ets.org/Media/Research/pdf/RR-03-07-Sinharay.pdf

‧ 國

立 政 治 大 學

N a tio na

l C h engchi U ni ve rs it y

98

Terrell, D., (1996). Incorporating monotonicity and concavity conditions in flexible functional forms, Journal of Applied Econometrics, 11, 179-194.

Zhang, X., (1999). A Monte Carlo study on the finite sample properties of the Gibbs sampling method for a stochastic frontier model, Journal of Productivity Analysis, 14, 71-83.

相關文件