Capacity Building in Macroeconomic Modelling for Africa: the Nigerian case

1
Capacity Building in Macroeconomic Modelling for
Africa the Nigerian case

• Jacques Kibambe

2
Objectives

• Design a prototype model for small, open and
  developing African economies
• Framework (data permitting)
• Real sector supply demand
• Monetary sector
• Public sector
• Foreign sector
• Application forecasting and policy analysis
• Update and clean databases relevant for the
  Nigeria-specific model
• Develop a customised and country-specific
  macro-econometric models for the Nigerian economy
  
• Document the results
• Train researchers on applying the model

3
The challenges

• How should macro models be structured to serve
  the unique characteristics of African economies?
• How can sophisticated theoretical structures and
  advanced econometric techniques essentially
  designed for developed economies be adopted or
  customized to be applicable and effective within
  the African context?
• How are we going to deal with the impediments to
  ensure that macro models are useful and credible
  tools?

Macroeconomic modelling in Africa
4
The process

• Modelling involves the following steps
• How should the structure look like (economic
  theory)?
• How can the theory be implemented (general model
  specification)?
• How do the data look like (analysis of data
  properties)?
• Data defects
• Structural breaks
• Outliers
• Missing observations
• Data inconsistencies (adding up constraints,
  definitions)

Macroeconomic modelling in Africa
5
The process

• What is the value of the parameters in the model
  (estimation)?
• Are the results statistically adequate
  (hypothesis tests )? Is the model theoretically
  adequate ( smell test or do the results make
  economic sense)?
• Can the model predict (out of sample simulation)?
• Are the models response characteristics
  reasonable (signs and orders of magnitude of
  model multipliers)?
• This sequence of steps does not constitute a
  one-shot process
• Rather, if at any stage the implied hypothesis is
  rejected, the model should return to an earlier
  stage and revisit the theory, use new data,
  revise estimates, etc. until the model survives
  all stages.

Macroeconomic modelling in Africa
6
The structure

• Supply side capturing the long-run structure
• Supply side integration with demand side price
  adjustment mechanisms
• Policy effects monetary and public sectors
• Impact of the rest of the world Trade sectors
  BOP

Macroeconomic modelling in Africa
7
The sectors

• Real sector
• Supply side
• Demand side
• Price adjustment mechanisms
• Monetary sector
• Public/Government sector
• Trade and BOP sector

The structure of macro models
8
From Theory to Practice

• Steps (Theory Practical application)
• Specification
• Data analysis
• Estimation
• Simulation
• Forecasting
• Policy evaluation

The methodology
9
Macro models in the African context

• Economic structure of African countries?
• Small open developing economies (low income?)
• Supply-side constrained
• Inadequate production capacity
• Labour market inflexibilities
• Investment constraints (domestic and foreign)
• Important role of agriculture (primary sectors)
• Parallel markets (formal informal sectors)
• Dual currencies (multiple currencies)
• Price controls
• Dual labour markets (skilled unskilled)
• Transition and structural change

10
A core African model the ideal specification

• Some measure of potential output

• Some measure of output

11
A core African model the ideal specification

• Expenditure components

• Foreign sector

12
A core African model the ideal specification

• Prices (cpi or gdp deflator)

13
A core African model overriding principles

• Stability and robustness vs. detail
• Full model properties (dynamic response
  characteristics of model) vs. single-equation
  specification and performance

14
The Nigerian case data dictates

• Data sources
• International Financial Statistics published by
  the International Monetary Fund
• World Development Indicators published by the
  World Bank
• Statistical Bulletin of the Central Bank of
  Nigeria
• Data problems
• Unavailability
• Unreliability
• Structural breaks dummies

The Nigerian case
15
The Nigerian case data dictates

• Proxies and replacements
• Inflation ?gdpdeflator and not cpi
• gdpresidual gdpreported gdpcalculated
• Proxy for ucc (1prime/100)exchange rate (due
  to unavailability of tx and ilong)
• oildisease Poilexchange rateOilprod
  (inequitable income distribution effects of oil
  production negative bias towards oil production
  industry)

The Nigerian case
16
The Nigerian case data dictates

• Dummies
• Managed Dummy variable to denote the period of
  managed exchange rates (1994 to 1998 1)
• Milrule Dummy variable to denote the period of
  military rule (1983 to 1998 1)
• Oildum Dummy variable to denote the effects of
  1973 oil crisis (1973 and 1974 1)
• Transdum Dummy variable to denote the transition
  from military to civilian rule (1994 to 2001 1)

The Nigerian case
17
Model specification a demand-side approach

• GDP identity (Keynesian)
• Real household consumption expenditure
• Real gross fixed capital formation

The Nigerian case
18
Model specification a demand-side approach

• Real exports of goods and services
• Real imports of goods and services
• Domestic prices GDP deflator

The Nigerian case
19
Model specification a demand-side approach

• Production price index
• Naira/Dollar exchange rate
• Nominal prime lending rate

The Nigerian case
20
Model specification a demand-side approach

• User-cost-of-capital
• Oil disease
• Inflation

The Nigerian case
21
Model specification a demand-side approach

• Openness
• Gross domestic expenditure

The Nigerian case
22
Estimation results

• Stochastic equations
• GDP deflator
• Production price index
• Nominal prime lending rate
• Naira/Dollar exchange rate
• Real exports of goods and services
• Real imports of goods and services
• Real household consumption
• Real fixed gross capital formation

The Nigerian case
23
Estimation results GDP deflator

• ?LDEF87 0.1741TRANSDUM - 0.2911OILDUM
  0.1243 - 0.4186(LDEF87 0.8757LPZ-(1-
  0.8757)LPPI87-LGDE_GDP -0.7191)

The Nigerian case
24
Estimation results Production price index

• LPPI87 0.1199LUCC 0.2704LOILPD 3.6335

The Nigerian case
25
Estimation results Prime lending rate

• LNPRIME 0.8883LNREDISR 0.0023INFL 0.5142

The Nigerian case
26
Estimation results Naira/Dollar exchange rate

• LEXND 1.4964LOG(DEF87/USDEF87) -
  0.0487LOG(NGDP/(NUSGDP))-1.2591MANAGED
  0.4925MILRULE

The Nigerian case
27
Estimation results Real exports of gs

• ?LRXGS 0.2792?LRUSGDPN 0.7883?LOILPD
  -0.4371(LRXGS 0.2458LRUSGDPN - 0.4365LOILPD
  - 3.34358)

The Nigerian case
28
Estimation results Real imports of gs

• ?LRZGS 1.4444?LRGDP 1.1461?LRELP
  0.1698?LRZGS(-1) -1.0353(LRZGS 1.6057LRGDP -
  1.5689LRGDPMILRULE - 1.0822LRELP
  18.7713MILRULE 13.1481)

The Nigerian case
29
Estimation results Real household consumption

• ?LRCONS - 0.1304?LOILDISEASE
  1.1264?(LNMSA/LDEF87) 0.0894 - 0.6491(LRCONS
  - LRGDP - 0.2517MILRULE 0.5404)

The Nigerian case
30
Estimation results Real fixed investment

• ?LRIFIX 1.6451?LRGDP - 0.1754?LEXND
  -0.8555(LRIFIX - 1.4077LRGDP 0.2453LUCC -
  0.4087LOPEN 0.2546MILRULE 0.3571TRANSDUM
  6.006)

The Nigerian case
31
Dynamic model properties the baseline

• Model closure
• Identities
• Definitions
• In-sample simulation (1971 2001) allow for lag
  structures
• Add-factor calibration
• In-sample 2000 - 2001
• Fine-tune the model with latest updated actual
  figures

The Nigerian case
32
Generating a forecast

• Identify the set of exogenous variables
• Forecast assumptions
• Global outlook
• National policy
• Determine ex ante (forecast) values for exogenous
  variables in model for the forecast period
• Run the ex ante simulation

33
Generating a forecast

• gdpUS
• PUS
• Pz

• Set of exogenous variables
• g
• inv
• M
• irediscount
• Poil(dollar)
• Oilprod
• Pworld(dollar)

34
Generating a policy scenario

• Identify the relevant exogenous and policy
  variables
• Generate new shocked values for the policy
  variables
• Substitute the baseline (or actual) variables
  with the shocks in the model
• Run the simulation (ex post or ex ante)
• Compute percentage differences between the
  baseline and shocked values of the response
  variables

35
4 Policy scenarios
36
Project outputs

• Models
• Estimated
• Calibrated with-in sample
• Baseline ex ante simulation (last value of
  exogenous variables taken forward)
• Documented results
• Training
• Software
• Model application forecast policy analysis
• Exogenous variable forecasting techniques