The Impact of OSHA and EPA Regulation on Productivity

The Impact of OSHA and EPA Regulation on Productivity

 Author：
Wayne B. Gray
Source ：
NBER Working Paper No. 1405 (1984/07)

先說感想：

1. 這樣一篇直指EPA與OHSA副作用的文章，在台灣可能不會出現=>因為學界不少需要靠公家單位的計畫經費來養活自己，政府單位的官員不會心胸大到願意提供數據給唱反調的學者.....而拿了公家單位經費的學者也不會笨到說金主不想聽的壞話.
2. 這是篇計量經濟的文章，用多元迴歸的方式來估算環安法規對於業界生產力的衝擊程度，不是用投入產出間的均衡聯立方程式組來計算；落落長58頁的working paper，許多地方也沒弄不清楚作者想表達的意義。
3. 環安法規的目的當然是改善環境品質與勞工安全衛生水準，從這篇研究看來：越嚴苛繁雜但無法落實的法規，不但沒達成以上目標，還必須對產業生產力衰退負責？！(但只須負0.5%左右的責任..實在有點微不足道..)
4. 文章當中提到EPA與OHSA成效不彰的原因在於：查核次數太少、罰款處罰規定太輕=>這點在台灣也一樣，不過老美因為動輒打官司，所以或許公司會比較守法；至於在台灣，人民比較不喜上法院/勞工比較少爆料去捅，所以事業單位的環安落實程度會更差？
5. 這個我想做的研究，這個也叫Wayne的老美在25年前就做了.......

Abstract：
This paper presents estimates of the impact of OSHA and EPA regulation on productivity. Production information for 450 manufacturing industries from 1958 to 1980 is merged with measures of regulation, including both information on compliance expenditures by industry and enforcement efforts by OSHA and EPA. Industries that faced higher regulation during the 1970s had significantly lower productivity growth, and a greater productivity slowdown, than industries that faced lower regulation. Under certain assumptions, the regulation is estimated to have reduced average industry productivity growth by .57 percent per year, 39 percent of the average productivity slowdown. These results are robust to variations in the model and the inclusion of other productivity determinants, including poor output growth and dependence on energy. The results also suggest a one-time cost of adjustment to regulation, so the long-run impact may be less than that estimated here. Both OSHA and EPA are found to target their enforcement effort towards those industries that are doing poorly in meeting the goals of the regulation. However, in the only area where benefits from regulation can be examined, worker injury rates and OSHA safety inspections, no significant benefits are found.


Download Link：
Here
隱含的邏輯假設：
增加的法規造成企業經營的限制與不確定，進而增加營運的成本與抑制企業在生產上的投資及成長
Regulation commonly imposes constraints on the firm's choice of production processes.
Additionally, the presence of regulation and the possibility of future changes in regulation, increases the uncertainty faced by firms, which could reduce investment in new rthods of production and inhibit
productivity growth.
理由：
Both EPA and 0SHA are required by law not to consider compliance costs when deciding what standards to propose. The standards proposed often specify the control techniques to be used, and tend to require capital—intensive compliance methods. The standards have also changed over time, in response to new information on health risks, political pressures from various groups, and court
decisions.

驗證方法：
由於難以直接衡量企業為符合法規所需增加的支出成本，故只能：
方法1
很單純的對照全產業法規干預前後的生產力成長率是否有顯著的差異
方法2
對照不同法規干預程度的不同產業，其生產力成長率是否有顯著差異
備註：
還是沒有辦法完全排除外生變數的干擾因素：例如不同時期的生產力成長差異、不同產業的差異
以這個case而言，米國的生產力成長率在1970s以後減小、難以歸咎原因是因為EPA與OHSA在1970年代成立後制定出一堆法規，拖累抑制產業成長率...當時時空背景的日本競爭力提升、石油危機或許才是主因

研究的難處：
There is no longer a way to separate the impact of regulation into measurement and actual effects, but it is possible to examine the magnitude of the impact of regulation on changes in measured productivity growth.
We could attempt to model directly the impact of regulation on the demands for and marginal products of different inputs, using a rre detailed nrdel of the production process, with impacts on productivity being derived as a result.
As the model becomes nore involved, it becomes ucre sensitive to possible misspecification or errors in the variables. Since compliance costs tend to be poorly measured, and are often not measured at all, this can cause problems for the nore complex models.
如何區別出當時石油危機所造成的生產力衰退等外部環境干擾，進而估算出政府環安法規所造成的拖累(程度) ？
一些學者對於環安法規乃至於各項因素對生產力衰退相關性與貢獻度的正反研究，可參見page12-13
Data & Sample:
Production information for 450 manufacturing industries from 1958 to 1980 is merged with measures of regulation, including both information on compliance expenditures by industry and enforcement efforts by OSHA and EPA

Research Design:
被解釋變數：這段時間內，450個產業的生產力指標與成長性
table 3.1 & 3.2
包含幾種不同算法的生產力成長指標：1973年前與後有明顯的差距
生產力=output/input , 生產力的微分=生產力成長率
1973年前與後生產力有明顯差距的解釋：
石油危機引發的經濟衰退、失業

解釋變數：環保工安法規的稽核與造成的支出增加
table 2.1 包含：
1973-1980 OHSA各年度的稽核次數、查核合法率、罰款金額、預估的安衛資本支出(Worker Health and Safety Capital Expenditures)
1973-1980 EPA各年度的稽核次數、查核合法率、環保資本與費用(Pollution Abatement Capital Expenditures and Operating Costs)
但對於各項環安支出的數據由抽樣調查而來，樣本可能不具代表性....


研究方法及過程結果：
所有解釋與被解釋變數的相關矩陣結果在page 27 table 5.2
(晚生20年太幸福了=>因為統計軟體進步太多，這個相關係數矩陣連顯著性都沒標出來)
精華在 page 29 table 5.3
不同的模式放不同的解釋變數進行迴歸，然後看解釋力與係數存在顯著性

由多元線性迴歸，假設各變數是屬於常態分配，原始數據當中可能有離群點，所以作者在page 33 的table 6.1  又去除離群點與用無母數的方式進行各變數間的相關係數檢定與迴歸
用spearmen 相關係數檢定而非pearson 相關(這點我在研討會上也被一位彰師大的博士電過.....)

page 37 table 6.3 考慮不同產業的能源與資本密集度的差異
所有變數都放進去回歸，解釋的R-Square竟然只有0.061=6%？！
其中環安因素的解釋力大約只有0.5 %

作者還做了許多後續的分析與檢定，但我已經消化不良，(我現在可以了解口試委員對我碩士論文的感受.....=>作了很多/內容很豐富沒錯，但核心是什麼？)


研究目的：
1.了解環安法規是否對生產力造成衝擊(Yes，檢定結果顯著)
2.估計衝擊程度並與先前相關研究進行比對(對生產力成長的拖累估計有0.57%，先前Denison的估計大約僅有0.35%)


感想
原來這就是working paper而不是正式的稿件
優點：
作者的思考脈絡與來龍去脈交代得很清楚
缺點：
當然就是落落長，把不是很關鍵的嘗試結果都show 出來，增加讀者的負擔