Galí 在結構式向量自我回歸 (SVAR) 中使用長期限制來辨認技術衝擊的影響 , 而後許多學者也嘗試使用此架構作為研究方法 , 其中部分學者的討論著重在每人平均工時是否具有單根。 然而技術衝擊的來源包含了許多面向 , 於是我們參考 Fisher (2006) 的做法 , 將技術衝擊的來源拆解為投資衝擊與中性技術衝擊。 我們運用投資的實際價格來捕捉投資衝擊 , 而投資價格會依據 Gordon(1989) 與 Greenwood,Hercowitz, and Krusell (1997) 所規範的類別做計算。 於是本文運用 SVAR模型來分析 , 探討投資衝擊與中性技術衝擊對於每人平均勞動工時的影響。
在雙變量 SVAR 中 , 當假設工時具有單根時 , 面臨正的技術衝擊時 , 工時在短期會下降 ; 當假設工時不具有單根時 , 面臨正的技術衝擊時 , 工時會產生正向的反應 , 且到長期也不會收斂。 此結果與過去學者研究研究一致。 然而技術衝擊的來源不再來自單一來源時 , 模型在投資衝擊的加入後變為三變量 SVAR 。 當假設工時具有單根時 , 面臨一個正的投資衝擊 , 工時會產生負向的反應 , 但不顯著 ; 而面臨正的中性技術衝擊時 , 工時仍在短期會顯著的下降。 以及當假設工時不具有單根時 , 面臨一個正的投資衝擊 , 工時會產生正向的反應 , 且到長期不會收斂 ; 面臨正面的中性技術衝擊時 , 工時會產生正向的反應 , 且到長期也不會收斂。 但再加入投資衝擊後 , 不管是否具有單根 , 三變量系統中的中性技術衝擊與雙變量的技術衝擊相比 , 都會使工時增添部分負面的影響。
透過不同 λ 的設定 , 對工時進行 HP-flter 濾波後 , 在 λ = 1600 時 , 不管在三變量系統 , 還是雙變量系統 , 結果都與假設工時具有單根時一致。 在 λ = 16000 時 , 分析結果會有較大的差異 , 對於台灣工時的處理 λ = 1600 的設定較適合。 且在台灣目前現有的資料上 , 我們以單根的設定來研究技術衝擊與工時的關係是相對合理的 , 所以當我們面臨正的技術衝擊時 , 工時是下降的。 在部分產業上工時對勞動的薪資有很大的影響 , 故需要調整合理的資源分配 , 避免擴大貧富差距。

Galí used long-term constraints in structural vector autoregression (SVAR) to identify technological shocks, and then many scholars have since tried to use this framework as a research method, Some of the scholars’ discussions focus on whether the working hours has a single root. However, the source of technology shock contains many aspects, so we refer to Fisher (2006) to Disassemble technology shock into " Investment-specifc Shock " and "Neutral Shock. " Wecapture investment-specifc shock by using the actual price of investment, which is calculated according to the categories specifed by Gordon (1989) and Greenwood, Hercowitz, and Krusell (1997) . Therefore, this paper uses the SVAR model to analyze the impact of investment-specifc shock and neutral shock on the working hours.
In bivariate SVAR model. First, working hours are assumed to have a single root. When faced with a positive technological shock, working hours decline in the short run. And then it is assumed that the working hours do not have a single root. When faced with a positive technological shock, working hours will have a positive response, and it will not converge in the long run. This result consistents with previous research by scholars. However, the source of technology shock is no longer from a single source, the model becomes a three-variable SVAR afer adding in investment -specifc shock. It is assumed that the working hours have a single root. When facing a positive investment-specifc shock, the working hours will have a negative but insignifcant response. When facing a positive neutral shock, the working hours will still decrease signifcantly in the short term. And then it is assumed that the working hours do not have a single root. When facing a positive investment-specifc shock,the working hours will have a positive response,and it will not converge in the long run. When facing a positive neutral shock, the working hours will have a positive response, and it won’t converge in the long run. After adding the investment-specifc shock, regardless of whether there is a single root, the neutral shock in the three-variable system will add some negative effects to the working hours compared with the two-variable technical shock.
We use HP-filter on working hours by different λ settings. When setting λ = 1600 , the results are consistent with the assumption that the working hours have a single root no matter in the three-variable system or the two-variable system. When setting λ = 16000,there will be a big difference in the analysis results. Working hours in Taiwan may be more suitable for the setting of λ = 1600. In addition, based on the existing data in Taiwan, it is relatively reasonable for us to study the relationship between technological shock and working hours with a single root setting. According to Taiwan’s data, when we face a positive technological shock, the working hours decline. Working hours in some industries have a great impact on the wages of labor, so it is necessary to adjust a reasonable allocation of resources to avoid widening the gap between the rich and the poor.

Christiano, Lawrence, Martin Eichenbaum, and Robert Vigfusson.(2003)“What Happens afer a Technology Shock?” NBER Working Paper No. 9819.
Dedola, L., and Neri, S. (2007).“What does a technology shock do? A VAR analysis with model-based sign restrictions.” Journal of Monetary Economics, 54(2), 512-549.
Fisher,Jonas. (2006). “The Dynamic Effects of Neutral and Investment-Specifc Technology Shocks.” Journal of Political Economy,114, 413–52.
Francis,N.,andRamey,V.A.(2005). “Is the technology-driven real business cycle hypothes is dead? Shocks and aggregate fuctuations revisited.” Journal of Monetary Economics,52(8), 1379-1399.
Francis, N., and Ramey, V. A. (2009). “Measures of per capita hours and their implications for the technology-hours debate.” Journal of Money, credit and Banking, 41(6), 1071-1097.
Galí, Jordi. (1999). “Technology, Employment, and the Business Cycle: Do Technology Shocks Explain Aggregate Fluctuations.”American Economic Review, 89, 249–71.
Gordon, R. J. (2007). “The measurement of durable goods prices.” University of Chicago Press.
Greenwood, J., Hercowitz, Z., and Krusell, P.(1997). “Long-run implications of investment-specifc technological change.” Te American economic review, 342-362.
Kydland, Finn E.,and Prescott, E. C.(1982). “Time to build and aggregate fuctuations.” Econometrica: Journal of the Econometric Society, 1345-1370.
Uhlig, H. (2004). “Do technology shocks lead to a fall in total hours worked?.” Journal of the European Economic Association, 2(2-3), 361-371.