Estimation of time-series supply and use tables to support analysis of productivity growth in Malaysia

Transcription

1 Estimation of time-series supply and use tables to support analysis of productivity growth in Malaysia M Yusof Saari a, Mazlina Shafii b and Rosmiza Rosly b a Department of Economics, Universiti Putra Malaysia a Institute of Agricultural and Food Policy Studies, Universiti Putra Malaysia b Malaysia Productivity Corporation The Fourth Asia KLEMS Conference, 31 July 01 August, 2017, Hitotsubashi Summer Institute, Tokyo, Japan

2 Presentation Outline Objectives Estimation Approach Results of Estimation Sources of Output Growth

3 1 Objectives This paper has two objectives To estimate time-series supply and use tables (SUTs) in Malaysia for the periods Estimated by applying so-called SUT-RAS approach To estimate sources of output growth by using the estimated timeseries SUTs - Estimated by applying KLEMS methodology

4 Estimation Approach 2 - Structure of estimated SUT Commodities m Industries n Imports Value added Commodities Industries Final demand Total m n 1 2 k output V (n m) U (m n) E m k S m (1 n) w (1 n) Total input S G G V = supply matrix, V ij represents the amount of commodity j produced by industry i (includes imports) U = matrix of intermediate demand, U ij represents the amount of commodity i used by industry j E = matrix of commodity deliveries to k categories of final demands m = vector of industry import w = vector of industry value added S = S = vector of commodity gross output G = G = vector of industry total output

5 2 Estimation Approach The six imperatives for successful overall group strategy - Structure of estimated SUT Commodities m Industries n Imports Value added Commodities Industries Final demand Total m n 1 2 k output V (n m) U (m n) E m k S m (1 n) w (1 n) Total input S G G Three benchmark SUTs had been used: 2000 SUTs 2005 SUTs 2010 SUTs After harmonization for all three tables: 73 industries/commodities for periods industries/commodities for periods industries/commodities for all periods 73 industries/commodities 82 industries/commodities

6 2 Estimation Approach - RAS-SUT approach Euro Method Estimated SUTs based on growth rates of (i) value added, (ii) total final demand and (iii) total imports Two important assumptions: (i) the shares of industries in the production of commodities remain constant and (ii) the fixed input coefficients determine the relations of all product inputs to production of industries EU KLEMS method Estimated SUTs based on (i) output, (ii) intermediate input, (iii) total final demand and (iii) total imports Shares of industries in the production of commodities remain constant in the supply table for the projection periods SUT-RAS WIOD method SUT-RAS which has been applied for the estimation of world input-output database (WIOD), is a theory-based that minimizes the deviations of the projected SUT structure from the benchmark period Joint estimation of SUTs and does not require the availability of SUTs by commodities, which are derived endogenously

7 2 Estimation Approach - RAS-SUT approach Results of updating Spanish SUTs at basic price MAPE Rank WAPE Rank SWAD Rank Make matrix (72 x 72) EURO EUKLEMS SUT-RAS Supply table (make matrix + imports, 72 x 73) EURO EUKLEMS SUT-RAS Total intermediate use ( 72 x 72) EURO EUKLEMS SUT-RAS Total final demand (72 x 3) EURO EUKLMES SUT-RAS SUT-RAS method greatly improves the Euro and EUKLEMS methods in estimating Spanish SUTs Sources: Temurshoev, U and Timmer, MP (2011) Joint estimation of supply and use tables Papers in Regional Science, 90, MAPE = Mean Absolute Percentage Error; WAPE = Weighted Absolute Percentage Error; SWAD = Standardized Weighted Absolute Difference

8 2 Estimation Approach - RAS-SUT approach Commodities m Industries n Imports Value added Commodities Industries Final demand Total m n 1 2 k output V (n m) - RAS - U (m n) - RAS - m (1 n) w (1 n) Total input S G E m k S G V = supply matrix, V ij represents the amount of commodity j produced by industry i (includes imports) U = matrix of intermediate demand, U ij represents the amount of commodity i used by industry j E = matrix of commodity deliveries to k categories of final demands m = vector of industry import w = vector of industry value added S = S = vector of commodity gross output G = G = vector of industry total output Time-series components of final demand, imports, value added and total output are compiled from the survey data The intermediate demand and the supply table are adjusted biproportionally by means of the RAS method

9 3 Results of Estimation - Sensitivity analysis The accuracy of the estimated SUTs depending largely on the quality of data and estimation procedures To assess the relative performance of our methodology, we compare our estimates in 2010 with the actual benchmark SUTs in 2010 i Mean absolute percentage error (Butterfield and Mules, 1980): MAPE= 1 mn m n i=1 j=1 x ij xij true 100 true x ij ii Weighted absolute percentage error (Mínguez et al, 2009): WAPE= m n i=1 j=1 true x ij k l x true kl true x ij x ij true 100 x ij iii Standardized weighted absolute difference (Lahr, 2001): SWAD= m i=1 n j=1 true x ij x true xij x ij k l x kl true ²

10 3 Results of Estimation - Sensitivity analysis Results of updating Malaysian SUTs for intermediate demand matrix A Results for Malaysian SUTs MAPE WAPE SWAD 1 Yearly benchmark estimates Fixed benchmark estimates Fixed benchmark estimates (known FD) B International comparison Spanish SUTs (Temurshoev and Timmer, 2011) Results clearly show the superiority of our approach compared to other two alternative approaches Our results are also comparable to international practices Yearly benchmark estimates Fixed benchmark estimates Fixed benchmark estimates (given known FD)

11 3 Results of Estimation - Sensitivity analysis Intermediate input coefficients, estimated vs actual in Correlation coefficient = Our methodology and find that our estimation provides good approximation for the true SUTs and comparable with the findings from other studies

12 EXPORT 2010 Import 2005 Export 2005 Results of Estimation 3 - Sensitivity analysis Import and export structures, estimated vs actual 005 Import Export SUT SUT Import Export 2010 Import SUT SUT

13 Sources of Output Growth 4 - Main broad economic sectors Applying standard growth accounting The production functions are assumed to be separable in these inputs as the starting point: y j = g j (Y ij ) = f j (K j, L j, X j, T) where Y is output, K is an index of capital service flow, L is an index of labour service flows and X is an index of intermediate inputs Standard growth accounting decomposition of output growth into the contribution of each input and TFP (denoted by A Y ): ln Y jt = v X jt ln X jt + v K jt ln K jt + v L Y jt ln L jt + ln A jt where the contribution of each input is defined as the product of the input s growth rate and its two period average share

14 Sources of Output Growth 4 - Main broad economic sectors Main findings Agriculture, Construction and services sectors largely driven by the improvement in TFP Mining and Quarrying sector is mainly driven by capital investment (in particular mineral exploration) Manufacturing sector is largely driven by intermediate inputs contributed by outsourcings and trade fragmentation activities Improvement in the high-skilled labor productivity that in line with the current 11MP Agriculture and Construction Educational mismatch must be carefully addressed!

15 Sources of Output Growth 4 - Main broad economic sectors Educational mismatch for selected occupations (%) Clerical support workers 131 Over-qualified 869 Technician and assoc professionals Over-qualified Professionals Under-qualified Qualified Mismatch Source: Household Income Survey (2010) In the case of high educational mismatch, occupation categories are more appropriate to represent skills than education entertainments

16 Sources of Output Growth 4 - Priority sectors Main findings Chemical & Chemical Product capital investment is the major contributor to the output growth Machinery TFP and intermediate inputs are the main drivers for output growth Electrical & Optical Equipment indicates negative growth which largely contributed by intermediate inputs Wholesale Trade and Retail Trade TFP and intermediate input are the main drivers for output growth

17 What s next Development of inter-regional input-output tables A joint work between the following agencies Universiti Putra Malaysia Integrated Sustainability Analysis (ISA), University of Sydney Malaysia Productivity Corporation Methodologies are based on the Industrial Ecology Virtual Laboratory (IELab), developed by ISA 2017 Project kick-off and data collection Productivity analysis at regional level is expected to be completed by 3 rd quarter of Jun Jan Mar Jun Sept Dec Database expected to be completed by 1 st quarter of 2018

18