Eurostat Macro integration

Presented by Piet Verbiest Statistics Netherlands

Macro integration Reconciliation of inconsistent statistical data on a high level of aggregation Balancing is reconciling inconsistent statistical information from independent sources brought together in an ‘accounting’ framework consisting of well-defined variables, accounting identities on combinations of variables and other less strict relations between the sets of variables.

Macro integration National accounts an example

National accounts Comprehensive overview of all economic transactions in a country Quarterly and annual report of a country Key indicators  Gross domestic product (GDP): economic growth;  Gross national income  Consumption of households, investment, foreign trade  Government debt  Employment

7 Labour accounts National accounts in the Netherlands Supply and use tables Sector accounts

Supply and use tables Variables and basic identities identities (1) P + M = IC + C + I + E (2) Y = P - IC (3) Y = C + I + E - M (4) Y = W + OS/MI 8

What we want: 9

What we get: 10

Macro integration / balancing

Macro integration / balancing

Macro integration / balancing

14 SUPPLYUSE Output of industries Import Total Input of industries Cons Export Invest. Total Commodities Y Total PM IC+YCEI Value added = PIC+Y = GDP

15 SUPPLY USE Output of industries ImportTotal Input of industries Cons. Export Invest. total Commodities Y TotalPMIC+YECI P - IC = Y = GDP  P–IC = Y =C+I+E-M PMICCIE++++=

16  Commodities: 500  Industries: 150  Final expenditure: 20  Simultaneous: cup and cop

17

Eurostat Macro integration

Presented by Jacco Daalmans

Mathematical models 2+9=10 5=7 15/2=7 22=17 1=0 3+7=10 6=6 22=17+5 Mathematical Models =25

Mathematical models Can be automated Reproducible results Flexible Large scale applications

BUT: Small discrepancies, without known cause

Example 1: Whisky Imports = Consumption Given: Imports = 5, Consumption=0 Model outcome could be: Imports= 2.5 Consumption = 2.5 NOT DESIRABLE!

Example 2: Remaining discrepancies Production (P) = 930 Imports (M) = 275 Interm. Cons. (IC)= 450 Cons. Invest. Export (CIE)= 740 P+ M = IC + CIE 1205 ≠ 1190 P – IC = CIE – M 480 ≠ 465

Example 2: Remaining discrepancies Production (P) = Imports (M) = Interm. Cons. (IC)= Cons. Invest. Export (CIE)= P+ M = IC + CIE 1205 ≠ =1200 P – IC = CIE – M 480 ≠ =473

Different models

STONE’s Method Broad applicability Achieves consistency by solving a minimum adjustment problem

STONE’s Method Searches for a result with minimum deviation from the input. Mathematical: Translation to a least squares optimization problem Consistency rules translate to constraints of the model.

STONE’s Method Linear constraints, like: Total is the sum of components: Manufacturing = Food + Textiles + Clothing; Commodity balances; Total use = Total supply; Definitions: Value added = Output – Intermediate consumption

Extensions Inequality constraints: Total Use ≥ 0 Soft constraints: Stocks of perishables goods ≈ 0 Ratio constraints: Value added Tax / Supply = 0.21 Refineries: use of crude oil / output ≈ 0.7

A man with a watch knows what time it is A man with two watches is never sure (Segal’s Law)

Reliability weights Important instrument to steer the results.

Example 2: Remaining discrepancies Production (P) = Imports (M) = Interm. Cons. (IC)= Cons. Invest. Export (CIE)= P+ M = IC + CIE 1200=1200 P – IC = CIE – M 473=473

Example 2: Remaining discrepancies Production (P) = Imports (M) = Interm. Cons. (IC)= Cons. Invest. Export (CIE)= P+ M = IC + CIE 1200= =1200 P – IC = CIE – M 473= = 480 green = p and IC more reliable

Conclusions Mathematical methods powerful instrument Elaborate modelling constructions possible But should be used properly!