1

THE OXFORD GLOBAL ECONOMIC MODEL

AN OVERVIEW

February 2011

Oxford Economics

Abbey House

THE OXFORD GLOBAL

ECONOMIC MODEL

AN OVERVIEW

April 2012

2

The Oxford Global Economic Model an overview

CONTENTS

INTRODUCTION 3

THEORETICAL MOTIVATIONS 4

COVERAGE OF THE OXFORD MODEL 5

THE CORE MODELS 6

An outline of the Oxford country models 6

The Oxford Global Model structure 7

Simulations 12

Conclusions 15

EXTENSIONS TO THE CAPITAL FLOWS MODEL 21

Interest rates 21

Credit conditions 26

Simulations 30

SECTOR BREAKDOWN 32

Value added 32

Employment 33

ANNEX A: COUNTRIES COVERED IN THE OXFORD GLOBAL ECONOMIC MODEL 34

ANNEX B: TECHNICAL STRUCTURE OF THE OXFORD GLOBAL ECONOMIC MODEL 35

ANNEX C: A SCHEMATIC MODEL 37

3

The Oxford Global Economic Model - an overview

Introduction The Oxford Global Economic Model has been constantly evolving over the past 30 years or

so, reflecting a continuous programme of enhancement, in response to the changing issues

facing the economy, policy-makers, companies and other institutions. Most recently, we have

introduced two sets of major changes: to the way in which the Model handles international

capital flows and the impact of credit conditions and to the linkages with major sectors. This

updated Overview covers these changes as well as a more general description of the Model

and its main characteristics.

It has long been one of Oxford Economics’ guiding principles that many of the most

important and interesting macroeconomic issues are inherently international. Globalization

means that policy makers and analysts have to form judgements about important economic

developments not only in their own country, but in their major trading partners as well. A

change in US monetary policy, for instance, has repercussions for the whole world; oil and

commodity price shocks have been the major source of terms of trade movements in Europe

in the last quarter century or so; governments are increasingly collaborating over monetary,

fiscal and environmental policies. All of this means that single country econometric models,

which treat world trade, world prices and exchange rates as exogenous, are not best suited to

analysing some of the most important issues of interest to financial and business economists.

The root cause of this integration is the massive increase in trade and capital flows between

countries in the post-war period, and Oxford Economics’ client base is testament to the

growth in interest in international issues. With offices throughout the world, in the UK,

elsewhere in Europe, the US and Asia, Oxford Economics aims to combine access to local

information and expertise with a global outlook to provide a truly international service. The

Oxford Global Economic Model reflects that priority, as coverage of the major trading

countries has both deepened, and widened.

The Oxford Model improves on previous vintages by incorporating well-behaved, theory-

consistent models for all of the individual countries covered, not just the big seven. It

maintains the tradition of allowing for significant cross-country differences in model

structures, but ensures that those differences truly reflect economic, as opposed to economic

model-builders’, idiosyncrasies. Where possible, and it is possible in the majority of cases,

4

the functional form for equations is left the same across countries. Parameters differ of

course, and this means that different countries exhibit different behaviour in response to

shocks (although economy structure also accounts for variations). Now, however, tracing

the root cause of these differences, and attributing them to underlying behaviour or

structure, is much simpler. For instance, real wage rigidity is higher in some countries than

others, and specific coefficients in wage and price equations reflect this. Unemployment will

tend to rise further and faster in these countries in response to an adverse demand shock,

even though the functional form of wage and price equations is identical across countries.

Theoretical Motivations Different types of models suit different purposes. The days of relying on a single, large

macroeconometric model as the definitive “pictorial” representation of an economy are gone.

However, the same demands which drove the construction of large-scale models of the 1970s

and 1980s are still there: business economists still need to forecast, they still need to analyse

the effects of government policy, and they still need to study the implications of different

theories about behaviour.

Broadly speaking, there are three types of model designed to help the business economist in

these tasks. At one extreme, there are the purely statistical models known as vector

autoregressions (VARs). Their strengths are short-term forecasting (usually six months to a

year or so) and the generation of stylised facts. However, they are much less useful for

longer-term forecasting and, because they lack any economic structure, they cannot be used

for policy analysis.

At the other extreme are the so-called computable general equilibrium models (CGEMs).

These models’ equations are derived by assuming private agents solve dynamic optimisation

problems, and they typically do not have error terms, or residuals, like econometrically-

estimated relationships. They are calibrated so that in equilibrium they reproduce historical

averages of key macro variables. Their strength is their high degree of rigour, but when

econometricians perform statistical tests on them, they typically do badly relative to the

traditional models.

At Oxford Economics we take the third approach. However, we recognise that both the

approaches described above have important lessons for traditional model-builders. A good

test of a macro model is whether it does as well as a VAR in reproducing short-run

5

behaviour, and whether its long-run relationships are supported by cointegration in VARs.

Also CGEMs have taught us the importance of theory, and that it is often better to impose a

coefficient to match a tried-and-tested stylised fact than to stick slavishly to coefficients

estimated from short samples of data. The main advantage of the macroeconometric

approach is that it provides both a forecasting tool and a tool for policy analysis. This

approach is the closest we will get to the “jack-of-all-trades”, combining sensible forecasts

with well-founded analysis.

Coverage of the Oxford Model The ‘core’ Oxford Global Model now comprises forty-four country models together with

headline indicators for another 33. There are also six trading blocs to complete the world

coverage. The country models are fully interlinked via trade, prices, exchange rates and

interest rates, with the blocs completing all the world coverage.

The models can be classified into five groups1:

I II III IV V

US Sweden Poland Denmark Eastern Europe

Japan Switzerland Hungary Finland Latin America

Germany Belgium Russia Norway Africa

France Netherlands Czech Republic Ireland OPEC

Italy Spain Brazil Portugal Rest of OECD

UK Austria Argentina Bulgaria Rest of World

Canada Mexico Chile Croatia

China Australia South Africa Greece

South Korea Indonesia Romania

Taiwan Malaysia Slovakia

Hong Kong Philippines

Turkey

Singapore

Thailand

India

Typical number of variables:

400+ 250-400 150-250 75-150

6

In addition, the model includes a bloc of world variables such as oil and commodity prices,

world GDP and industrial production, OECD average inflation, aggregates covering the

Eurozone group etc. The country models (I-IV) are identical in structure but the bigger

models incorporate greater disaggregation and more financial sector detail. The blocs

identify the key aggregates - GDP, consumer prices, exchange rate and current account - for a

further 33 countries (see Annex A for full list of countries covered).

The core models

An outline of the Oxford country models

The structure of each of the country models is based on the income-expenditure accounting

framework. However, the models have a coherent treatment of supply. In the long run,

each of the economies behaves like the textbook description of a one sector economy under

Cobb-Douglas technology in equilibrium. Countries have a natural growth rate, which is

ultimately beyond the power of governments to alter, and is the result of population and

productivity growth. Output cycles around a deterministic trend, so at any point in time we

can define the level of potential output, corresponding to which is a natural rate of

unemployment. Firms are assumed to set prices given output and the capital stock, but the

labour market is imperfectly competitive. Firms bargain with workers over wages, but they

get to choose the level of employment. Countries with high real wages get high

unemployment in the long run, and countries with rigid real wages get persistently high

unemployment relative to the natural rate.

Inflation is a monetary phenomenon in the long run. All the models have vertical Phillips

curves, so expansionary demand policies put upward pressure on inflation. Unchecked,

these pressures would cause the price level to accelerate away without bound, and in order

to prevent this we have endogenised monetary policy. For the main advanced economies,

the latter is summarised in an inflation target, and interest rates are assumed to move up

whenever inflation is above the target rate, and/or output is above potential (a so-called

‘Taylor rule’). The coefficients in the interest rate reaction function, as well as the inflation

target itself, reflect our perceptions of how hawkish different countries are about inflation. A

by-product of this new system is that simulations under fixed interest rates make sense for

only a couple of years or so. If you do not “do” monetary policy, and Phillips curves are

1 New models are currently being tested for additional countries.

7

vertical, then you end up with hyperinflation (or hyperdeflation, depending on the shock)

after a few years.

Quantitative Easing (QE), whereby the central bank prints money and uses it to purchase

assets in order to stimulate the economy, has played an important role as a policy tool in the

aftermath of the 2008/09 financial crisis. This is included in the model as an exogenous

variable for the US, Japan, the Eurozone and the UK. QE tends to lower government bond

yields and boost share prices via portfolio effects.

Demand is modelled in much the same way as before. Consumption is a function of real

incomes, real financial wealth, real interest rates and inflation. Investment equations are

influenced by “q-theories”, in which the investment rate is determined by its opportunity

cost, after taking taxes and allowances into account. Countries are assumed to be “small”, in

the sense that exports are determined by demand and a country cannot ultimately determine

its own terms of trade. Consequently, exports are a function of world demand and the real

exchange rate, and the world trade matrix ensures adding-up consistency across countries.

Imports are determined by real domestic demand and competitiveness.

There is a trade-off between detail and tractability. In general, our approach has been to

aggregate where it is not clear that disaggregation (i) improves the quality of forecasts or

analysis or (ii) serves particular users’ needs. From a practical point of view, aggregation

tends to make it easier to identify the model with theoretical counterparts, and thus gives us

a clearer idea of its relative strengths and weaknesses. Many financial flows have been

aggregated, and government accounting conventions have been standardised at a relatively

high level of aggregation. On the other hand, we continue to disaggregate the components

of personal income, the categories of investment and the energy sector, partly because we

believe that doing so helps us to forecast better, but also because we recognise that these

variables are of interest to particular users. Annex C presents a schematic summary of a

typical country model.

The Oxford Global Model structure

Model variables are divided into demand and supply, core and non-core. Coverage of core

variables is standard across all country models; non-core coverage is determined by data

availability and country-specific requirements. Core demand variables include all the

aggregate expenditure components, at constant and current prices, monetary policy variables

8

and financial variables. The demand non-core includes disaggregated consumption and

investment, as well as important indicator variables such as retail sales and car sales. Core

supply consists of variables determining the natural levels of output, unemployment and

real wages. Prices are also disaggregated in the core supply block. Non-core supply

disaggregates employment and nominal earnings. Separate blocks build up the government,

personal and corporate sector flow accounts, while the G7 energy model is also included as a

distinct entity in some versions.

The following sections describe the structure and theoretical motivation of some of the key

equations in the core model. Tables 1-5 present the responses of the model to shocks to

illustrate the properties of these equations. (Annex B explains the technical structure of the

Oxford Model equations.)

Consumption

We follow the standard econometric treatment pioneered by Hendry et al (1985). The

equations take the form:

∆c = a1 * ∆y + a2 * ∆u -a3 * (c(-1) - a4 * y(-1) - (1-a4) * W(-1) + a5 * R(-1))

where lower case letters denote logs and c, y and u are consumption, real income and

unemployment respectively, while W and R refer to the financial wealth-income ratio and

real interest rates. All the variables that the modern treatments stress - with the exception of

human wealth which is difficult to quantify - are included in our formulation; real interest

rates, taxes and wealth are what matter. In addition, these error-correction formulations

appear to mimic consumption smoothing in a number of countries very well.

Investment

Three aspects of gross fixed investment are identified in the Oxford Model: private business,

private housing and government, which is exogenous.

The equations for business investment are based on so-called q-theories of investment. In

these, capital is time-consuming to install and these adjustment costs drive a wedge between

the post-tax marginal product of capital and its marginal cost. Profit maximising firms

invest when the marginal return is greater than the replacement cost (q > 1), and reduce

9

investment, or even scrap, when the reverse holds. In the long run, the capital stock reaches

its desired level, all investment is replacement, q = 1 and the familiar marginal productivity

relationship holds. The equations are backward looking and take the following form:

∆i = a1 * q - a2 * (i(-1) - k(-1)) + a3 * ∆y

where i is private sector business fixed investment, k is the equivalent capital stock and y is

GDP; q is defined as the post tax marginal product of capital relative to the real interest rate.

With Cobb-Douglas, constant return to scale technology, the capital-output ratio is constant

in the long run, and equal to the post-tax, post-depreciation real interest rate divided by the

capital share. There are also short-term accelerator effects from changes in output, which can

be justified in a q-framework if some companies are credit-constrained.

Personal sector housing investment is determined analogously to consumption, by real

income, wealth and interest rates, since it is considered part of a portfolio of spending

decisions taken by households.

International Trade

Trade flows are disaggregated into fuel, non-fuel goods, and services. The non-fuel goods

components reflect the bulk of exports and imports for most countries and we focus on those

here. Exports and imports are demand determined:

∆x = ∆wt - a1 * cu - a2 * ∆wcr - a3 * (x(-1) - wt(-1) - a4 * trx)

∆m = b1 * ∆tfe + b2 * ∆wcr - b3 * (m(-1) - tfe(-1) -b4 * wcr(-1) - b5 * cu(-1))

x refers to exports of non-fuel goods; m to the equivalent imports; wt is world trade; tfe, total

final expenditure; wcr, relative unit labour costs; and cu, capacity utilisation as measured by

model estimates of the output gap. The time trends capture secular shifts in a country’s

world trade share caused by non-price factors, and the impact of the long-term increase in

the specialisation of production on import penetration. Trade competitiveness elasticities are

typically between 0.3 and 0.6; and most country models satisfy the Marshall-Lerner

conditions, so that a sustained improvement in competitiveness will lead to an improvement

in the trade balance in the long run.

10

The equations for trade in services are analogous to those for non-fuel goods, while imports

of fuel meet the gap between, on the one hand, domestic and export demand and, on the

other, domestic production. All trade prices are a weighted average of domestic and world

prices.

Linkages between economies

Thus, the Model links the individual countries in a number of ways:

• Trade (Exports driven by weighted matrix of trading partners’ import demand)

• Competitiveness (IMF relative unit labour costs where available, relative prices

elsewhere)

• Interest Rates and Exchange Rates

• Commodity Prices (e.g. oil, gas and coal prices depend on supply/demand balance;

metal prices depend on growth in industry output)

• World Price of Manufactured Goods

Energy

Oil (& gas/coal) prices are determined by the interaction of demand and supply in the global

market. The following drivers are key:

• World demand (+: impact on price)

• World supply (-)

• Oil reserves (-)

• Oil stocks (-)

• Spare capacity (+): the gap between max oil output and actual output

World supply is the aggregate output of oil in all oil producing countries. Demand for oil is

estimated by adding up the oil demand of the main economies. Oil demand is linked to

economic growth and relative oil prices – fully integrated with the rest of the Oxford Model,

while Saudi Arabia is assumed to act as swing producer all through the model

11

Carbon emissions

Carbon emissions are calculated at the domestic level, as a function of domestic energy

consumption, and aggregated globally.

Core Supply

Given its importance to overall model properties, this is probably best summarised as a

block, rather than equation by equation. The following diagram is a useful, if simplistic,

description of the key features of the model’s supply side:

The north-east quadrant shows the production function with diminishing returns, relating

output to employment. Tangents to the production function are the marginal product of

labour, which in equilibrium equals the real wage. These tangents trace out a demand for

labour in the south east quadrant - our employment equation (nd). Given a fixed labour

supply (ns), the intersection generates the equilibrium real wage consistent with no

involuntary unemployment and normal (or potential) output. The latter is traced out along

the vertical aggregate supply curve (ASl) in the northwest quadrant.

In the short run, however, relatively rigid real wages generate involuntary unemployment

(nd ≠ ns), while nominal inertia means that the short run relationship between real wages and

the price level is shown by the hyperbola in the south-west quadrant. Short run changes in

labour demand then trace out a positively sloped short run aggregated supply curve (ASs),

Employment

Real Wage

nd

ns

ASl

AD

Output AS

s

12

ensuring that changes in aggregate demand (AD), as derived from an IS-LM system,

translate into short run changes in prices and output, although the long run effects are felt on

prices alone.

In short, the employment equation defines a level of real unit labour costs (real

wages/productivity) which is constant in the long run. Consistent with this level of real unit

labour costs are natural levels of output and unemployment. When the economy is away

from these natural levels, inflation and interest rates move to bring the economy back

towards equilibrium. The larger nominal and real rigidities are, the larger and longer-lived

real disequilibria.

Algebraically, the employment equation solves in the long run for the constant level of real

unit labour costs, given by labour’s share in the production function, while the wage and

price equations solve in the long run for the level of unemployment consistent with this

labour share. In the short run, both wage and price equations incorporate nominal and real

wage rigidity, which ensure the existence of “involuntary” unemployment and monetary

effects on the real economy.

With vertical Phillips and aggregate supply curves, monetary policy determines the inflation

rate, while structural, or supply side policy determines the unemployment rate. The NAIRU

(non-accelerating inflation rate of unemployment) is related to the so-called ‘tax wedge’ (the

gap between the total real cost of labour to employers, including social security

contributions, and the real value of post-tax wages received by employees), and to real

energy prices.

Simulations

Below we present simulations of monetary and real shocks. But before we plunge into the

details, it is worth describing what we would expect to see, in general terms, in the different

types of simulations. These expectations, which are based on textbook analyses, help to

explain why we have imposed the theoretical restrictions described above.

The first distinction to draw is that between real and nominal variables. In general, in the

long run only real shocks should affect real variables, like GDP and unemployment.

Monetary shocks will change nominal variables, like the price level and nominal wages, but

not the ratio of the two - real wages.

Price Level

13

Two factors complicate the picture, however. First, monetary shocks do have real effects in

the short to medium run because of the presence of nominal and real rigidities in wage and

price setting. Moreover, the greater these rigidities, the longer it takes for the model to reach

equilibrium following a shock, and 10 years, in many cases, does not constitute the long run.

Second, endogenous monetary policy means that, while real variables may respond in

similar ways across countries, nominal variables need not. Our interest rate reaction

functions ensure that inflation is stabilised, but how long that takes to happen depends on

the size of nominal rigidities and on the credibility of the monetary authorities, as

summarised in the parameters of the reaction function. Consequently, long run impacts on

the price level, nominal earnings, the exchange rate etc. can differ substantially across

countries. In addition, the combination of nominal rigidities and a rule targeting inflation

causes the model to be cyclical at business cycle frequencies, so that in many cases it will not

have settled down even ten years after a shock. This is apparent in all the simulations below.

Tables 1-5 summarise the following simulations for US, Japan, UK, China, India, Brazil and

the Eurozone

1. Fiscal shock - Government consumption raised by 1% of GDP

2. Investment up 1% of GDP ex ante

3. Monetary shock - Interest rates up 1% point

4. Monetary shock - 5% exchange rate depreciation against the $US

5. World oil price + $10pb

Note that the simulations are run for all the countries and not only the one concerned.

1. Fiscal shock - Government consumption raised by 1% of GDP (Table 1)

Table 1 shows the effects of a sustained rise in government expenditure on goods equivalent

to 1% of GDP. The key points to note are:

• The rise in demand leads to a prolonged rise in output.

• However, with potential output unaffected directly by such a ‘demand’ shock,

inflationary pressures quickly emerge. This in turn leads to higher interest rates, which

squeeze private sector expenditure.

14

• In the long-run, output returns close to base levels - i.e. to potential output. Inflation also

returns to base levels, but the price level and the nominal interest rates remain

permanently higher, as does the real exchange rate - it is these responses which embody

the ‘crowding out’ mechanism.

• The deterioration in competitiveness means that the current account position is

permanently worse, as is the government deficit. And the government budget position

continues to deteriorate throughout the period of the simulation as higher borrowing

raises debt servicing costs.

It is worth highlighting that the long-run effect of such a demand shock on output is, if

anything, likely to be negative in the Oxford Model because of its impact on business

investment. Typically, investment rises in the short term, reflecting the accelerator effects of

higher demand. In the long-run, however, the effect of higher real interest rates dominates

so that investment falls below base levels. This in turn will lead over time to a lower capital

stock and hence lower potential output.

2. Investment up 1% of GDP ex ante (Table 2)

This shock is in many ways analogous to the fiscal shock presented in Table 1. However,

because the higher investment adds to the capital stock, and hence potential output, it leads

to a sustained rise in output and a better inflation-output trade-off than higher government

consumption.

3. Interest rates up 1% point (Table 3)

This simulation involves a sustained ex post rise in interest rates, with monetary policy

assumed not to respond to the consequential changes in output and inflation. As noted

earlier, such a policy would not be sustainable in the long run; we therefore present results

only for two years.

This simulation implies that, as a ready reckoner, each 1% point rise in interest rates reduces

GDP growth in the US and the Eurozone by about ½% point in its first year, while inflation is

reduced by about ½% point after two years.

15

5. Monetary shock - 5% exchange rate depreciation (Table 4)

The improvement in competitiveness caused by the depreciation boosts net trade in the short

term, and hence GDP rises above base although this positive effect may be mitigated by

weaker real consumption caused by rising import prices. Both prices and earnings gradually

rise - eventually by the full extent of the depreciation.

6. World oil price plus $10pb (Table 5)

All of the simulations presented so far represent nominal shocks. A rise in the world oil

price, in contrast, represents a ‘real’ shock. Higher energy costs lower the profitability of

production and therefore reduce firms incentives to supply, cutting potential output. As a

consequence, this shock leads to a sustained loss of GDP.

Conclusions

This overview has outlined the Oxford Model of the world economy and illustrated its key

simulation properties. It has shown that, while the Model exhibits ‘Keynesian’ features in

the short to medium term, its long-run properties are ‘neo-classical’ - i.e. attempts to raise

growth and employment by boosting demand will ultimately lead to higher prices, with

output in the long run determined by supply-side factors - productivity and population

growth.

Oxford Economics is continually working to improve the model. Attention has focussed

recently on adding more detail to the financial sector and its relationship with the real

economy. Comments and suggestions for further analysis are, of course, very welcome.

16

Table 1: Fiscal Shock - Government Consumption Raised By 1% of GDP (% changes from base unless otherwise stated)

Year US Japan UK China India Brazil EuroZone

GDP 1 1.1 1.1 0.8 1.2 1.0 1.0 0.6

3 0.7 1.4 0.8 1.2 0.7 1.1 0.4

5 0.4 0.2 0.6 0.5 0.9 0.8 0.3

CPI 1 0.1 0.2 0.0 0.1 -0.1 -0.1 -0.1

3 1.0 1.5 0.9 1.2 0.1 -0.1 0.5

5 2.6 3.3 2.0 2.9 0.4 0.7 1.8

ER 1 0.5 0.2 0.6 0.1 0.4 0.3 0.3

3 1.6 1.7 2.2 1.3 0.3 0.3 0.9

5 3.1 3.4 3.4 2.6 0.7 1.1 2.2

ET 1 0.5 0.3 0.3 0.3 0.3 0.5 0.2

3 0.3 0.6 0.5 0.4 0.4 0.8 0.2

5 0.1 0.2 0.3 0.3 0.5 0.6 0.1

BCU% 1 -0.4 -0.3 -0.4 -0.3 -0.5 -0.2 -0.3

3 -0.7 -0.6 -0.6 -0.6 -0.6 -0.3 -0.3

5 -0.8 -0.5 -0.7 -0.4 -0.8 -0.5 -0.3

RSH 1 0.7 0.4 0.7 0.0 0.0 0.3 0.4

3 0.7 1.4 1.1 0.5 0.7 0.8 0.8

5 0.6 1.7 1.0 0.9 0.4 1.2 1.2

17

Table 2: Investment up 1% of GDP ex ante (% changes from base unless otherwise stated)

Year US Japan UK China India Brazil EuroZone

GDP 1 1.0 1.2 0.8 1.0 1.0 1.0 0.7

3 0.9 2.0 1.3 1.1 1.3 1.4 0.8

5 1.4 1.2 1.8 0.7 1.7 1.2 1.1

CPI 1 0.1 0.0 0.0 0.1 0.0 -0.1 -0.1

3 1.1 0.1 1.0 1.0 0.1 -0.2 0.4

5 2.0 0.3 2.3 2.5 0.4 0.6 1.3

ER 1 0.5 0.2 0.6 0.1 0.4 0.2 0.3

3 1.7 1.9 2.5 1.1 0.7 0.4 1.1

5 3.1 4.2 4.2 2.3 1.0 1.2 2.3

ET 1 0.4 0.4 0.3 0.3 0.3 0.5 0.2

3 0.4 0.7 0.7 0.4 0.7 0.9 0.2

5 0.5 0.4 0.8 0.4 0.9 0.8 0.2

BCU% 1 -0.3 -0.4 -0.4 -0.3 -0.9 -0.2 -0.4

3 -0.6 -0.8 -0.8 -0.5 -1.3 -0.3 -0.3

5 -1.0 -0.9 -1.2 -0.4 -1.4 -0.3 -0.4

RSH 1 0.6 0.4 0.7 0.0 0.1 0.2 0.4

3 0.4 1.3 1.1 0.5 1.2 0.4 0.6

5 0.0 1.8 1.1 0.7 1.0 0.5 0.7

18

Table 3: Monetary Shock - Interest Rates Up 1% Point (% changes from base unless otherwise stated)

Year US Japan UK China India Brazil EuroZone

GDP 1 -0.4 -0.1 -0.3 -0.2 -0.3 -0.4 -0.4

2 -1.0 -0.4 -1.2 -0.6 -0.4 -0.8 -0.8

CPI 1 -0.1 -0.2 -0.2 0.0 -0.4 -0.4 -0.2

2 -0.5 -0.5 -0.4 -0.1 -0.7 -0.5 -0.6

ER 1 -0.2 0.0 0.0 0.0 -0.2 -0.3 -0.2

2 -0.9 -0.2 -0.6 -0.1 -0.6 -0.8 -0.8

ET 1 -0.2 0.0 -0.1 -0.1 -0.1 -0.2 -0.1

2 -0.4 -0.1 -0.6 -0.2 -0.2 -0.4 -0.2

BCU% 1 0.1 -0.1 0.0 0.0 0.3 -0.1 0.0

2 0.0 -0.1 0.2 0.2 0.4 0.0 0.1

19

Table 4: Monetary Shock - 5% Exchange Rate Depreciation against the $US (% changes from base unless otherwise stated)

Year US Japan UK China India Brazil EuroZone

GDP 1 0.3 0.2 0.2 0.2 0.2 0.5 0.2

3 0.1 0.4 0.5 0.2 0.0 0.9 0.1

5 -0.2 -0.6 0.3 -0.2 0.0 0.3 0.0

CPI 1 0.2 0.4 0.5 0.6 1.2 1.3 0.4

3 0.7 1.8 1.4 1.6 2.7 2.4 1.3

5 1.3 3.4 2.4 2.8 3.9 3.9 2.3

ER 1 0.1 0.1 0.3 0.1 0.3 0.8 0.2

3 0.6 1.1 1.7 1.2 1.7 2.4 1.0

5 1.1 2.6 2.7 2.1 3.0 3.7 2.1

ET 1 0.1 0.1 0.0 0.0 0.0 0.2 0.0

3 0.0 0.1 0.3 0.1 0.1 0.4 0.0

5 -0.1 -0.3 0.1 0.0 0.0 0.2 0.0

BCU% 1 0.2 0.1 0.0 0.3 -0.5 0.3 0.1

3 0.2 0.5 0.4 0.1 -0.1 0.4 0.2

5 0.1 0.9 0.5 0.2 -0.1 0.3 0.2

RSH 1 0.3 0.2 0.4 0.2 1.3 0.4 0.8

3 0.3 0.8 1.0 0.5 1.3 0.1 0.8

5 0.1 1.2 0.8 0.6 0.5 0.1 0.8

20

Table 5: World Oil Price + $10 pb (% changes from base unless otherwise stated)

Year US Japan UK China India Brazil EuroZone

GDP 1 -0.1 -0.1 -0.1 -0.2 -0.3 -0.2 -0.2

3 -0.5 -0.4 -0.2 -0.6 -0.6 -0.5 -0.2

5 -0.4 -0.4 0.0 -0.2 -0.5 -0.5 -0.1

CPI 1 0.4 0.2 0.5 0.4 0.7 0.6 0.3

3 0.6 0.3 0.3 0.8 1.4 1.2 0.5

5 0.6 0.1 0.2 0.9 2.0 1.5 0.6

ER 1 0.1 0.0 0.1 0.1 0.1 0.2 0.0

3 0.1 0.0 -0.1 0.4 0.6 0.8 0.2

5 0.1 -0.1 0.0 0.5 1.3 1.1 0.3

ET 1 -0.1 0.0 0.0 -0.1 -0.1 -0.1 0.0

3 -0.2 -0.2 -0.1 -0.2 -0.3 -0.4 -0.1

5 -0.2 -0.1 0.0 -0.1 -0.3 -0.4 -0.1

BCU% 1 -0.2 -0.2 -0.1 -0.2 -0.2 -0.1 -0.2

3 -0.2 -0.2 -0.1 0.0 -0.2 -0.1 -0.2

5 -0.2 -0.1 -0.2 -0.1 -0.3 -0.1 -0.2

RSH 1 0.1 0.0 0.1 0.1 0.6 0.2 0.4

3 -0.1 0.0 -0.1 0.2 0.4 0.0 0.0

5 -0.1 -0.2 0.1 0.0 0.2 0.0 0.0

21

Extensions to the Capital Flows Model

The global financial crisis of 2008-09 highlighted some areas for enhancement in Oxford

Economics’ Global Economic Model. We have extended the coverage of financial variables

and strengthened the links between the financial and real spheres of the economy.

Interest rates

Policy and money market interest rates

For all countries, a policy rate has been added to the model that is based on a Taylor Rule.

The money market rate has been modelled as a mark-up on the policy rate, this mark-up

then affects bank lending rates and influences consumption and investment behaviour. In

addition, more detail on bank lending rates to consumers has been incorporated and

modelled as a mark-up on interbank rates.

Government bond yields, including credit ratings

We have estimated a model for government bond yields that incorporates some of the main

determinants of long term interest rates identified by the economic literature. The database

used for the estimation included data for 29 countries between 1996 and 2010. The

coefficients were estimated through a panel fixed effects estimator and the estimated

equation explains more than 90% of the variability of long-term interest rates. The

determinants of government bond yields considered were:

• A country-specific constant estimated through the panel cross-section fixed effects

estimator. The constant aims to capture the mostly time-invariant and country-specific

institutional factors that affect the average level of the yield in the period considered,

such as the quality and credibility of public institutions, the credit history of the country

and the structure of local financial markets.

• Nominal short-term interest rate. This captures the level of the yield curve and the

inflation rate. The coefficient suggests that an increase in the short-term rate of 100 basis

points causes long-term rates to rise by around 50 basis points.

• Benchmark government bond yields (US or Germany depending on the region). This

factor accounts for the correlation among yields of different countries. An increase in the

benchmark yields by 100 basis points causes long-term rates to rise by around 50 basis

points.

22

• Government debt as a percentage of GDP. We expect this variable to have a non-linear

effect on long-term interest rates. In particular, we expect the impact of an increase in

debt to be larger the higher the initial level of debt. All things equal, an increase of the

ratio of debt to GDP from 30% to 40% increases interest rates by approximately 5 basis

points while an increase from 130% to 140% increases rates by around 25 basis points.

• Government budget deficit. An increase in the budget deficit is commonly associated

with higher interest rates, as it signals acceleration in the future accumulation of debt. We

estimated that an increase of the deficit by 1% of GDP raises long-term yields by around

20 basis points. The result is broadly in line with previous estimates by the International

Monetary Fund and others and reflects the shared view in the literature that deficit has a

stronger impact on interest rates than debt.

• Credit Rating. We constructed an indicator of a country’s credit rating as a rescaled

average between 0 and 20 of the credit ratings from the three main agencies (where zero

is default and 20 is ‘AAA’). We estimated a non-linear relationship between credit rating

and government bond yields, such that the impact of a downgrading is larger the more a

country approaches default. We modelled the credit rating indicator as a linear

combination of the main factors taken into consideration by the rating agencies when

evaluating the ability of a government to repay their obligations, following similar

approaches in the economic literature (fiscal position, foreign reserves, GDP growth and

per capita levels, an indicator of government effectiveness). The equation explains two

thirds of the variability of the average credit rating assigned by rating agencies. In

particular, the rating indicator falls by 0.3 points on average if the government deficit

increases above 5% of GDP and by a further 1.4 points if it rises above 12% of GDP.

Moreover, the rating shrinks by 0.4 points if government debt rises above 70% of GDP

and by further 0.8 points if it rises above 130% of GDP.

• Inflation. Although inflation expectations are partly captured by the short-term rates, we

have also included an explicit term for CPI inflation, as the coefficient is significant and

has the correct sign. An increase in inflation by 1 percentage point causes long-term

yields to increase by around 20 basis points.

• Bond Stress is an exogenous variable that was not included in the estimation. In normal

conditions bond stress remains flat at 0, but it can be used to implement a country-

specific shock on long term interest rates.

23

• Quantitative Easing is included in the model for the US, the UK, Japan and the

Eurozone. It is an exogenous variable that feeds into the respective country’s bond yields

and the credit conditions measure (explained in detail in the next section). In the US, an

increase in QE by 10% of GDP lowers bond yields by 80 basis points in the first year,

while in the Eurozone the same shock causes a fall of 60 basis points in German bond

yields. Credit conditions in both cases ease considerably.

In the following chart we describe the relationships between the variables mentioned above:

Bond yields, credit ratings and macroeconomic variables

Corporate and consumer interest rates Corporate spreads measure the difference in the yield between corporate and Treasury

bonds with the same duration. In the financial economics literature, government bonds are

often referred to as “risk-free” assets and the spread represents the risk premium that

remunerates the investor for holding the risky corporate bond; the higher the perceived risk

in the asset, the larger the risk premium.

The risk of default perceived by investors depends on several macro and microeconomic

factors that need to be taken into account when trying to endogenise corporate spreads.

Moreover, it is necessary to distinguish between the long-term and short-term behaviour of

corporate spreads and, as a consequence, to identify the factors that determine the

equilibrium level and the short-term fluctuations of corporate spreads.

Government bond yieldsGovernment bond yields

Credit ratingCredit ratingGovernment budget deficitGovernment budget deficit

Government debt

Government debt

Short-term interest ratesShort-term

interest rates

Gross Domestic Product

Gross Domestic Product

External debtExternal debt

Foreign currency reserves

Foreign currency reserves

Political effectiveness

Political effectiveness

24

Several factors have been identified by the literature as the main determinants of corporate

spreads2:

• The Treasury yield curve: the slope of the yield curve is a predictor linked to current and

future macroeconomic conditions i.e. spreads are related to cyclical factors, which impact

upon perceived default risk.

• The level of interest rates affects firms’ net present value which, in turn, influences their

solvency and the perceived default risk.

• Aggregate measures of asset value, level of indebtedness of (non-financial) firms, and

stock and bond volatility which summarise firm-specific characteristics at a

macroeconomic level.

• Measures of stress for the banking sector and counterparty risk such as financial sector

debt, household debt, variations in house prices, unemployment and economic growth.

• Liquidity shocks can severely affect corporate spreads and a negative relationship

between the two factors has been well documented in the literature.

• Investors’ risk aversion can shift dramatically and for long periods in response to

macroeconomic and financial shocks, influencing both stock and bond markets.

Consequently, a proxy for risk aversion could be modelled to capture its impact on

corporate spreads.

Using data on corporate bond yields for the US3, Japan and Germany, we estimated the

following equation, where chpsh is the percentage change in the stock market from the

previous quarter.

rcorp – rlg = α + β*(rcorp-rlg)-1 + γ*chpsh

Comparing the estimated and actual corporate bond yields indicates that assuming that the

micro factors are constant over time is a reasonable approximation (Chart 1). The estimated

equation suggests that a 50% fall in equity prices raises corporate bond spreads by 100 basis

points (γ=0.02).

2 Krainer, J. (2004) “What determines the credit spread?”, FRBSF Economic Letter, No. 2004-36 and Christensen, J. (2008) “The corporate

bond credit spread puzzle”, FRBSF Economic Letter, No. 2008-10 3 The corporate bond yield for the US was constructed as a weighted average of the yields on AAA, AA, A and B, with the weights derived

from their share of the total market.

25

Chart 1: Estimated and actual corporate bond yields for the US

As with corporate bonds, the spread of consumer borrowing rates over the short-term money

market rate are fundamentally determined by the riskiness of the borrower to the lending

institution, which may vary over the economic cycle. But as consumers typically have no

option but to borrow from retail banks, the institutional characteristics of the banking sector

are also important.

The literature has identified a number of determinants of consumer borrowing rate spreads

which capture both of these aspects4:

• As with corporate bonds, current macroeconomic conditions as reflected in the Treasury

yield curve are a determinant of the spread.

• The mismatch in timing between demand for loans and supply of deposits requires retail

banks to hold some deposits in short term instruments, which are subject to interest rate

risk. This in turn pushes up the spread to compensate.

• Related to the second point, the amount of reserves banks hold to meet deposit

withdrawal demand and regulations on the level of banks’ capital are also positively

related to the spread of consumer borrowing rates.

• Low levels of competition within the banking system and between banks and non-

financial institutions also increase the spread, as does the productive efficiency of the

institution.

4 For a very good review of the literature see Gropp, R., C. K. Sorensen & J-D. Lichtenberger (2007), “The dynamics of bank spreads and

financial structure”, ECB working paper No. 714

0

1

2

3

4

5

6

7

8

9

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

%

Source : Oxford Economics

Estimated

corporate bond

yield

Corporate

bond yield

26

• Finally financial innovations over time, such as internet banking, tend to reduce the

spread other things being equal.

In a similar vein to corporate bond yield spreads, we have modelled the spread of consumer

borrowing rates over the short term (3-month) interbank money market rate across a range of

countries5 as a function of the macroeconomic cycle, the previous quarter’s spread and a

country-specific dummy variable. The country-specific constant reflects factors that are

relatively constant over short time periods e.g. the degree of competition within each

country’s banking system does not change significantly from quarter to quarter.

As expected, the economic cycle has a small but significant effect on consumer borrowing

spreads. This result is unsurprising and is in keeping with the estimates in the academic

literature in which the other factors which determine the spread, in particular the degree of

market power, are found to be much more important.

Credit conditions

Based on work by John Muellbauer6, a measure of overall credit conditions, i.e. conditions

not reflected in the interest rate, that impacts on access to credit have been incorporated in

the Oxford Model.

In the first instance, this focuses on the mortgage market and would introduce a theory-

consistent framework to decompose changes in house prices into the impact of:

• the underlying demand/supply balance

• real mortgage interest rates (incorporating the explicit modelling of house price

expectations formation)

• demographic change

• credit supply shifts

This approach distinguishes between countries where credit market shifts have had a big

impact on house prices (Spain, Ireland, US) and those where there was little or no impact

(Germany, Italy, Sweden). It therefore permits the analysis of the impacts in the housing

market due to:

• falls in demand (real incomes)

• impact of changes in house price expectations on real interest rates

5 As the data for consumer borrowing rates is better than for corporate bond yields, we have widened our sample to include the UK, US,

Germany, France, Greece, Canada, Portugal and Italy. 6 Aron & Muellbauer ‘Housing wealth, credit conditions and consumption’ Oxford (2006)

27

• negative shifts in credit supply

Credit conditions also impact on household consumption behaviour. We quantify the impact

of tighter credit conditions on private consumption. This is done in a similar way to the

quantification of the impact of tighter credit conditions on house prices. Having established

the degree of tightening in credit conditions consistent with trends in financial markets and

bank lending, we base the elasticity of consumption to changes in credit conditions on the

relationship identified by Aron and Muellbauer, allowing for differences in the importance

of bank lending across countries.

Finally, share prices of a country are also affected by credit conditions in that country.

Tighter credit conditions lead to lower share prices and vice versa.

Balance sheet

The financial positions of households and corporations can give a valuable insight into the

structural integrity of a countries economy, its future path of economic growth, potential

micro credit crunch and house price shock scenarios, and worthwhile investment

opportunities. The balance sheet readjustment instigated by the financial crisis was a

significant factor underlying the depth of the recession. The ability to assimilate balance

sheet data is therefore vital for forecasting future credit shock scenarios. The Oxford Model

incorporates these financial positions, including total debt and asset levels, the flows of

wealth, and revaluations.

Net financial wealth

The overall structure for households

and corporations is similar, with

financial and non-financial corporations

separately identified. Net wealth is

equal to gross wealth minus liabilities.

Gross wealth is formed from current

assets, plus acquisitions of new assets,

adjusted for the change in value from

the last period.

The structure of liabilities in the model

differs somewhat more between households and corporations. Corporate liabilities consist of

-150

-100

-50

0

50

100

150

2005 2006 2007 2008 2009 2010 2011 2012 2013

Greece: Sectoral net wealth% of GDP

Source : Oxford Economics

Non-financial

corporations

Households

Financial corporations

Forecast

28

bonds, equities and loans, while households only consist of loans. However loans are more

disaggregated for households, into mortgages, personal loans and miscellaneous loans (such

as store cards, credit cards etc).

The following sections give a brief explanation of how and for what purpose each sector has

been incorporated into the global macroeconomic model.

Households

The level of household debt played a significant role in triggering the 2009 financial crises, as

the rise in defaults during a period of negative equity housing undermined the real value of

bank assets. The Model now incorporates the net wealth of households, which is

disaggregated into current assets and liabilities, as well as flows of wealth. Debt levels are

further broken down into both long term debt (i.e. mortgages) and short term debt.

The net financial position of households has been incorporated into the model such that it

directly impacts upon on consumption. This allows for a more rigorous and endogenised

intertemporal analysis of consumer expenditure. Investment in private dwellings is a now a

function of, amongst other variables, net wealth.

The model structure of household balance sheets

Net wealthNet wealth

Financial markets

Government

bond yields

Equity

values

Consumer

interest rates

Financial markets

Government

bond yields

Government

bond yields

Equity

values

Equity

values

Consumer

interest rates

Consumer

interest rates

Mortgage loans Personal loansMiscellaneous

liabilities

Household liabilities

Mortgage loansMortgage loans Personal loansPersonal loansMiscellaneous

liabilities

Miscellaneous

liabilities

Household liabilitiesHousehold assets

Acquisition of

new assets

Revaluation of

current assets

Household assets

Acquisition of

new assets

Acquisition of

new assets

Revaluation of

current assets

Revaluation of

current assets

Household expenditure

ConsumptionInvestment in

residential dwellings

Household expenditure

ConsumptionConsumptionInvestment in

residential dwellings

Investment in

residential dwellings

House pricesDisposable

incomeUnemployment

Macroeconomic fundamentals

House pricesHouse pricesDisposable

income

Disposable

incomeUnemploymentUnemployment

Macroeconomic fundamentals

29

Household assets are now subject to revaluations due to movements in house prices,

equities, government bond yields. This key relationship allows for a more rigorous

estimation of house price and stock market shocks, and monetary policy changes. The impact

on both residential investment and consumption, and ultimately the wider economy, can be

estimated.

Corporate

The net wealth position of companies caused economic growth to drag through 2009 and

2010. Financial corporations attempted to reduce their financial exposure to further potential

defaults, and began to deleverage. The resulting flight to quality and restriction in credit

caused a sharp drop in investment and consumption.

The model now incorporates the financial position of both financial and non-financial

corporations. Total debt is disaggregated into secured and unsecured liabilities. Gross wealth

is a function of both new acquisitions of financial assets and revaluations of current assets.

The revaluation of current assets is driven by government bond yields and stock prices.

The model structure of corporate balance sheets

Net wealth of corporations is currently being used as a macroeconomic indicator. It is not

being used to mimic stress levels in the economy, but to reflect the levels of risk aversion,

deleveraging and credit crunch. It can therefore be taken into account when considering

various macroeconomic scenarios, future investment, capital flows, and interest rates.

Net wealthNet wealth

Financial markets

Government

bond yields

Equity

values

Corporate

interest rates

Financial markets

Government

bond yields

Government

bond yields

Equity

values

Equity

values

Corporate

interest rates

Corporate

interest rates

Corporate assets

Acquisition of new assets

Revaluation of current assets

Corporate assets

Acquisition of new assets

Acquisition of new assets

Revaluation of current assetsRevaluation of current assets

Corporate debt

BondsLoans

Corporate debt

BondsBondsLoansLoans

Corporate misc liabilities

Equities

Corporate misc liabilities

EquitiesEquities

Corporate liabilities

EquityDebt

Corporate liabilities

EquityEquityDebtDebt

30

Although the net wealth of corporations does not currently impact on other variables within

the model, it has the potential to be incorporated in a number of extensions.

The balance sheet additions as a whole give a useful insight into each countries debt

structure. This can be particularly useful for policy implications.

Simulations

Tables 6-8 summarise the following simulations for US, Japan, UK, China, India, Brazil and

the Eurozone (where applicable):

1. Credit conditions decrease by 0.08

2. Bond stress increases by 100 basis points

3. 10% decrease in government revenue in Italy

1. Credit conditions decrease by 0.08 (same as what we estimate happened during

the crisis)

This simulation represents a tightening of credit conditions under the assumption that

monetary policy remains fixed. A decline in the availability of credit squeezes consumer

spending as well as residential and business investment. These effects are reinforced via a

negative impact on house prices. Lower domestic demand causes a decline in consumer

prices and therefore average earnings, while the current account improves due to a fall in

imports.

2. Bond stress increases by 100 basis points

Monetary policy, in this simulation, is also assumed to be fixed. The increase in bond market

stress raises long-term rates. This leads to declines in investment causing output to fall.

3. 10% decline in government revenue

A decline in government revenue causes significant widening of the fiscal deficit leading to

downgrades in Italy’s sovereign credit rating. Government’s borrowing costs increase as

sovereign debt is considered more risky which has adverse consequences on growth.

31

Table 6: Credit conditions decrease by 0.8

% changes from base unless otherwise stated

Year US Japan UK EuroZone

GDP 1 -0.5 -0.7 -0.4 -0.4

2 -2.1 -2.1 -1.8 -1.1

CPI 1 0.0 -0.2 0.0 0.0

2 -0.4 -0.9 -0.5 -0.2

ER 1 -0.2 -0.1 -0.2 -0.1

2 -1.4 -0.7 -1.5 -0.7

ET 1 -0.2 -0.2 -0.1 -0.1

2 -0.9 -0.8 -0.7 -0.3

BCU% 1 0.1 0.2 0.2 0.1

(% of GDP) 2 0.5 0.5 0.8 0.4

Table 7: Bond Stress increases by 1 % changes from base unless otherwise stated

Year US Japan China UK India Brazil EuroZone

GDP 1 -0.1 0.0 -0.3 -0.2 -0.4 -0.1 -0.1

2 -0.6 -0.1 -0.6 -0.6 -0.7 -0.5 -0.3

CPI 1 0.0 0.0 0.0 0.0 0.0 0.0 0.0

2 -0.2 -0.1 -0.3 -0.3 -0.4 -0.2 -0.1

ER 1 0.0 0.0 0.0 -0.1 -0.1 0.0 0.0

2 -0.4 0.0 -0.2 -0.5 -0.5 -0.3 -0.2

ET 1 -0.1 0.0 -0.1 0.0 -0.1 0.0 0.0

2 -0.3 0.0 -0.2 -0.3 -0.3 -0.2 -0.1

BCU% 1 0.0 0.0 0.0 0.0 0.3 0.0 0.0

(% of GDP) 2 0.0 -0.1 0.0 0.1 0.6 -0.1 0.1

32

Table 8: 10% decline in government revenue % changes from base unless otherwise stated

Year Italy

GDP 1 0.0

3 -0.2

5 -0.4

CPI 1 0.0

3 0.0

5 -0.4

ER 1 0.0

3 -0.1

5 -0.6

ET 1 0.0

3 -0.1

5 -0.2

BCU% 1 0.0

(% of GDP) 3 0.0

5 0.2

RSH 1 0.0

(%) 3 0.0

5 -0.1

Sector breakdown We have also extended the Global Economic Model to include value added and employment

by sector. The breakdown varies slightly by country, depending on data availability. For

instance, for the European Union it consists of 14 sectors – agriculture and forestry,

extraction, manufacturing, utilities, construction, distribution services, hotels and catering,

transport and communications, financial services, business services, public administration,

education, health and other services. A few additional sectors such as entertainment, arts

and recreation and real estate have been included for the United States. For Asia, the

breakdown is in some cases slightly less detailed due to data availability.

Value added

The sector breakdown reflects the input-output structure of each economy. For each sector

we calculate the total demand for that sector as a weighted average of value added in other

33

sectors, with the weights taken from input-output tables. We then use total demand to

estimate the value added for that respective sector since in the long run (everything else

equal) value added and demand grow in line with each other. Value added is also affected

by competitiveness (measured by relative unit labour costs) to a degree that reflects the

international openness of each sector. A typical equation for value added is as follows:

GVAxxx = f (TDxxx, WCR, TRENDA, (GVAxxx (-1))

Where:

GVAxxx = Gross value added in sector xxx

TDxxx = Total demand for sector xxx

WCR = Relative unit labour costs

TRENDA = Time trend

Consistency between the value added and expenditure approaches of activity is ensured by

scaling value added in each sector up or down to obtain expenditure-based value added as

the sum of value added in the sectors.

Employment

Employment by sector is derived from value added in that sector and sector-specific

productivity trends. As in the case of value added, consistency between the total

employment forecast and employment in all sectors is achieved by scaling the sector

employment variables up or down.

34

Annex A: Countries covered in the Oxford Global Economic Model 1. Countries covered in detail US Spain Denmark Poland Japan Netherlands Finland Hungary Germany Belgium Norway Russia France Switzerland Ireland Czech Republic Italy Austria Portugal Argentina UK Sweden Bulgaria Brazil Canada Australia Croatia Chile China Mexico Greece Indonesia South Korea Romania Malaysia Taiwan Slovakia Philippines Hong Kong Singapore Thailand South Africa Turkey India

2. Other countries covered in trading blocs Rest of the OECD Africa Rest of World Iceland Egypt Israel Luxembourg Kenya Myanmar New Zealand Morocco Pakistan

Sudan Syria OPEC Tunisia Vietnam Algeria Uganda Bangladesh Iran Cameroon Iraq Nigeria Saudi Arabia Latin America Venezuela Bolivia Colombia Costa Rica Eastern Europe Dominican Republic Kazakstan Ecuador Ukraine Panama Paraguay Peru Uruguay

35

Annex B: Technical Structure of the Oxford Global Economic Model

The equations which make up the Oxford Global Model are set out in the various EQN files

(e.g. see UKEQNS.HLP for details of the UK model). These typically fall into two groups:

(i) Behavioural relationships - e.g. relating wages to prices, productivity and

unemployment

(ii) Technical relationships - e.g. the national income identities.

It is the behavioural relationships which represent the analytical content of the Oxford

Model. In general, these equations have a standard ‘error correction’ format (ie simple

control feedback loops), where:

∆Yt = α0∆Yt−1 + α1∆Xt + α2∆Xt-1 - β (Yt-1 - γXt-1) + Rt (1)

with: Y=dependent variable

X=explanatory variable(s)

R= residual

∆= first-difference operator

The term in parentheses in equation (1) represents the long-run relationship between X and

Y. That is, when the model has reached (static) equilibrium - so that ∆Yt = ∆Yt-1 = ∆Xt = ∆Xt-1

= 0 - then Yt = γXt. Note, if Y and X are expressed in logarithmic terms, this equation implies

that a 1% increase in X will lead eventually to a rise of γ% in Y (i.e. ‘γ’ represents the long

term elasticity of Y with respect to X). Economic theory is used to determine the appropriate

explanatory variables to include in X and also determine any restrictions on the value of γ

(e.g. in the context of an equation relating to wages and prices, static homogeneity would

imply that γ=1). Cointegration techniques are used to estimate this long term relationship.

Of course, economies are frequently out of equilibrium. The terms in ∆Y and ∆X in equation

(1) therefore seek to model the adjustment of Y back to its long term relationship with X (i.e.

the ‘dynamics’ of the equation). So, if there is a 1% sustained rise in X then:

36

-Y will rise immediately by α1%

-In the next quarter, Y will rise by [(1 +α0 - β) α1 + α2 + βγ]%, and so on until..

-.....Eventually, Y will rise by γ%, which represents the end of the adjustment

process

The speed with which Y adjusts to its long run relationship with X depends, in particular, on

the size of coefficient β. Note for equation (1) to be stable, β must lie between 0 and 1.

However, the closer β is to -1, the faster the equation will reach equilibrium following a

shock. For short term forecasts, it is important to understand the dynamics of the model

equations as the long-term properties. These are illustrated for some of the key equations in

the G6 models in tables 1-5 in the main text.

37

Annex C: A Schematic Model

The following is a highly condensed version of a typical Oxford country model. The idea is

to present the model's key equations in a relatively accessible fashion, so that key inter-

variable relationships can be seen clearly. We stress that this is just a small part of the model

template - they typically consist of more than 200 variables - however, these equations might

be thought of as defining the model's theoretical core. As such, the functional forms are

identical across all the countries covered.

The equations presented are all "long-run" relationships; i.e. they abstract from dynamics.

We adhere to the convention that lower case mnemonics denote logs of variables.

Demand goods market c = a1*pedy + (1-a1)*(penw-pc) - a2*rrh (consumption)

st = gdp + e1*time (inventory level)

mgnf = tfe + c1*wcr + c2*time (non-fuel imports)

xgnf = wt -d1*wcr + d2*time (non-fuel exports)

money market mon = b1*gdp + (1-b1)*(prnw-pc) - b2*RSH (real money balances)

RLG = b3*RSH + (1-b3)*RLG,US + b4*GGDBT/GDP! (long bond rate)

rxd = rxd(expected) + log(1+RSH,US/400) - log(1+RSH/400) + RISK (exchange rate)

38

Supply capital accumulation K = (1-DELTA)*K(-1) + IPNR 7 (capital stock)

IPNR = K(-1) + f1*QR + short run GDP effects (non-residential

investment)

RRH = f6*RSH + (1-f6)*RLG - 100*inflation (expected) (real interest rate)

labour market and the nairu LS = PART*POPW (labour supply)

part = f2*(er-pgdp) (participation rate)

nairu = f3*WEDGE8 (natural rate of

unemployment)

ESTAR = (1-NAIRU/100)*LS (natural employment

level)

yhat = α*estar + (1-α)*k(-1) + g1*trend (potential output)

cumod = gdp - yhat (output gap)

epr = gdp -er +pgdp (employment)

er = pgdp + gdp - epr - f4*(up - nairu) (average earnings) prices pgdp = er -gdp + epr + f5*cumod (GDP deflator)

pmgnf = h1*pgdp + h2*(wpmf+rxd) + (1-h1-h2)*(wpc+rxd) (import prices)

cpix = j1*pgdp + (1-j1)*pm (consumer prices)

Government Policy monetary ∆RCB = l1*(inflation - inflation(-1)) + l2*(inflation - target) + l3*cumod (‘Taylor’

rule)

7 delta is potentially endogenised as a function of the output gap; this is not the case in current versions of the

models 8 wedge is the (log) difference between the real product wage and real take-home pay, and consists of direct,

indirect and payroll taxes, as well as producer prices relative to consumer prices.

39

fiscal Government spending and major tax rates all currently exogenous Rest of the World WT = trade-weighted average of trading partners' imports (world trade)

WPMF = trade-weighted average of import prices (world prices)

WPC = weighted average of world non-fuel commodity prices

VARIABLE DEFINITIONS (not elsewhere specified)

pedy; real personal disposable income penw; personal sector net financial wealth pc; personal consumption deflator gdp; gross domestic product tfe; total final expenditure wcr; relative unit labour costs prnw; private sector net financial wealth ggdbt; government gross financial debt gdp!; nominal GDP qr; Tobin's "q" popw; population aged 16-64 trend; Solow residual.