Please enable JavaScript.

Coggle requires JavaScript to display documents.

IT Productivity (Papers (B&H1996 (Data (IS spending by large U.S.…

- - - - There is no productivity paradox for 1987-1991
      - Dollar for Dollar, spending on IT creates more value than spending on any other capital
    - - Y=F(K,L,Z,S)
      - ISUR
    - - IS spending by large U.S. firms, collected by IDG
      - Annual survey of IS Managers
      - 1987-1991
    - - IT Capital - good return, IT Staff - Huge return
      - Bigger payback on IT than on any non-IT capital
      - Decreasing marginal product over time
  - - - Direct and Indirect Effects
      - Can direct and indirect effects be separated?
      - Are these effects systematically different by industry?
      - Automate, informate and Transform
    - - Augmentation FF
      - Why exponential FF?
        
        Allows estimation of direct and indirect effects
        
        If k=0, no augmentation - can be verified
        
        Additively Separable
        
        technological progress is exponential, convex, it accumulates and compounds
      - Optimal Levels of L,K and Z
      - xtgls with industry specific AR1 and correlated heteroskedasticity
  - - - Q=F(IT, K, L)
      - 36 countries, 9 years
      - Time Fixed Effects
      - Country Fixed and Country Random Effects
    - - Overall insignificant IT
      - Split sample - Developed have significant IT. Developing have non-significant IT
      - Experience curves for IT - First build complementary capabilities in non-IT capital and labor capabilities before investing in IT
  - - - Improved intermediate inputs: convenience, timeliness, quality, vairety
      - More competitive supplier industry spillsover more
      - IOS share information about demand and supply with suppliers. Leads to automated, accurate and speedy order replenishment
      - Process monitoring, CAM, flex production - defect, precision, flexibility of output
    - - Use of input-output tables to measure linkage between the focal industry and its supplying industries
      - BLS does not deflate for quality improvements, so productivity in one industry can be overstated at the expense of another. eg: Computer supplier
      - AR1: Outputs are linked through Smooth Business Cycles
      - Heteroskedasticity: If Responses to business cycles do not occur with the same magnitude in each industry. Variances may also differ with time
      - Vji ne Mji. Pj = Pj + ej. mj = mj + ej. theta mi + thetamu CD:Sum(Proportion from j j's Z)
    - - Supplier IT investments are more than covered by own productivity increases. In addition, these translate into $66-$72 million in output increases for downstream industries
      - Price deflator over-estimates true deflator by 30% at mean Z level
      - Z: 0.06; S: 0.024;
      - Measurement error: mean of 29% mu = phi/theta; ej = mu*zj
  - - - Relative Industry concentration
      - EDI
    - - Y = F(K,L,Z,M,C)
      - Weight = Wij = Vij/Sum(Vij)
      - Ci = Customer IT weighted by proportion of transactions
      - Portion of customer spillover depends on relative industry concentration = cusconcn/consn
  - - - Propensity Score Matching
      - Switching Regression
- - - - Error, possibility variability in production
      - Something Missing
      - different ways inputs combine to produce output outside of the direct effects.
  - - - Inter-temporally stable input shares
        
        Use Time Fixed Effects
    - - Derived from the income accounting identity
  - - - Optimality condition
    - - Meaning: 1% change in slope.., curvature
    - - Constraints
        
        Symmetry
        
        Each product must have atleast 1 +ve AES
        
        Not a measure of curvature
      - Defn
        
        AESij = Eij/Sj
        
        Meaning: >0, <0
    - - Defn = (FjXj/Sum(FiXi])(AERij-AESii)
      - Features
        
        Not Symmetric
        
        Example: IT and non-IT Capital
        
        Is a 2-factor, 1 input measure. AES is a one factor, 1 input measure
        
        Measures relative input share, since d ln(XjPj/XiPi)/d ln(pj\pi) = 1-MESij
  - - - One year to another
      - Technology and expertise moves slowly
      - Scale of demand
    - - Response to economy level shocks
      - Industries have different sizes and production functions
        
        Wooldridge test for AR1
- - - - Deflators
      - Lags
      - Mis-measurement
        
        IT Measured under Finance or Operations
        
        Only Hardware data
      - Previous decade - too small a stock of IT
      - Inter-temporal shift in input shares
  - - - Firm IO aggregation
        
        Nataf 1948 shows that if firm production functions are additively separable in inputs, then aggregation is valid.
      - Sub-input aggregation
        
        Leontief 1947 showed that if the MRTS of a group of inputs does not change with a change in inputs not in the group, then this group can be aggregated