ANALYSIS Global Scenario Development Using Moody’s ... · Global National Forecast With Alternative Scenarios” by Sunayana Mehra, October 2012. changes arise because of shifts

Global Scenario Development Using Moody’s Analytics Country ModelsIntroduction

Moody’s Analytics develops economic alternative scenarios that are now a basic tool for regulators of financial institutions around the world such as the U.S. Federal Reserve and the Bank of England. Regulators are increasingly inquisitive about the details of how such scenarios are created. The objective of this document is to provide such information.

ANALYSIS

Prepared by

Petr [email protected] of Economic Research

Contact Us

Email [email protected]

U.S./Canada +1.866.275.3266

EMEA +44.20.7772.5454 (London) +420.224.222.929 (Prague)

Asia/Pacific +852.3551.3077

All Others +1.610.235.5299

Web www.economy.com www.moodysanalytics.com

MOODY’S ANALYTICS

2 February 2014

Global Scenario Development Using Moody’s Analytics Country Modelsby PeTr ZeMCIK

Moody’s Analytics develops economic alternative scenarios that are now a basic tool for regulators of financial institutions around the world such as the U.S. Federal Reserve and the Bank of England. Regulators are increasingly inquisitive about the details of how such scenarios are created. The

objective of this document is to provide such information.

Moody’s Analytics produces four types of forecasts. The first is a baseline forecast that captures our view of the most likely forecast scenario. The second group consists of stan-dardized scenarios. Every month, Moody’s Analytics updates six standard alternative scenarios for 55 countries and nine regional aggregates. These scenarios capture the main risks to the baseline outlook as determined by Moody’s Analytics. Custom scenarios form the third category. Given informa-tion about a nature of the custom scenario, Moody’s Analytics forecasts a large number of economic variables for chosen countries. Finally, Moody’s Analytics produces forecasts for alternative macroeconomic scenarios cre-ated by regulators who provide assumptions behind these scenarios in the form of select-ed series and a brief narrative. Examples are the Federal Reserve’s Comprehensive Capital Analysis Review scenarios and the Bank of England Anchor scenarios, designed by the Prudential Regulation Authority, which is now a part of the central bank. The regula-tory scenarios can be viewed as a special case of custom scenarios.

The focus of this paper is on the regula-tory scenarios. The illustrative examples will be the CCAR exercise from November 20131

1 Somewhat confusingly, the Federal Reserve refers to this scenario as “CCAR 2014,” and this notation will be used as a reference throughout the article.

and the PRA Anchor scenarios from 2013H2. In the PRA 2013H2 run, Moody’s Analytics produced forecasts for 19 countries and two regions. Our extension of the CCAR 2014 macroeconomic scenarios was more com-prehensive, as we generated forecasts for 36 countries and nine regional aggregates.

Regulatory scenariosThe regulatory scenarios contain param-

eters that describe the primary risks to the economic outlook as perceived by the regu-lators. Parameters or assumptions in 2013 focused on high interest rates because of the tapering of the Federal Reserve’s quantita-tive easing program and the potential re-escalation of the dual banking and sovereign debt crisis in the euro zone.2 For example, the PRA published its baseline forecast and two versions of a stress scenario, one including a low path for interest rates and the other a high path.3

The Federal Reserve provided three scenarios for the CCAR exercise in 2013—Baseline, Adverse and Severely Adverse. The Baseline scenario described an outlook based on the prevailing consensus outlook. The Ad-

2 http://www.federalreserve.gov/bankinforeg/stress-tests/supervisory-baseline-adverse-and-severely-adverse-scenar-ios.htm

3 http://www.bankofengland.co.uk/pra/Pages/supervision/activities/anchorscenario.aspx

verse scenario was a credit-themed projec-tion, including relatively high interest rates. The Severely Adverse scenario contained a euro zone crisis.

Regulators provide time series for a number of variables, accompanied by a brief narrative. We broaden this narrative in or-der to forecast a full set of macroeconomic and financial variables for each country. An example of the U.K. narrative for the PRA 2013H2 low interest rate scenario is given in the Appendix 1. PRA provided time series for 14 variables for the U.K. and the U.S. and two euro zone series with its Anchor scenarios. The Federal Reserve published 16 U.S. vari-ables, plus three series for the euro zone, the U.K., Japan, and developing Asia for its CCAR scenarios (see Table 1).

Moody’s Analytics modelsMoody’s Analytics uses a large simulta-

neous equation model to generate a fore-cast for the U.S. and for each of 55 other countries. In total, these 56 countries cover more than 90% of global output. The U.S. model contains more than 1,600 variables. A small core set of variables is used as the basis for the Moody’s Analytics 55 other-country models. We will refer to our country model for the U.K. as an example of our country-model template. The U.K. example contains more than 100 variables (see Ap-

MOODY’S ANALYTICS

3 February 2014

pendix 2). We use the same basic structure for all countries, although specifications of equations may be altered to accom-modate idiosyncrasies of various countries. We also employ a wide range of financial market and credit risk models to create projections for variables specific to banks and other institutions.4 These models use macroeconomic forecasts under alternative assumptions as inputs. Finally, several satel-lite support models are used in the scenario development process.

Simultaneous equation modelEconometric forecasting models can be

loosely divided into three main types. The first is a purely data-driven model. Essen-tially, no theory is required to formulate and estimate such a model—for example a sys-tem of equations in a vector auto-regressive model. Though this model provides reason-able short-term forecasts, it does not give a user any insight into links among economic variables and is somewhat restricted with respect to the number of variables that can be included in the system. On the other

4 The time series from macroeconomic scenarios are used as inputs in a wide range of Moody’s Analytics models in areas such as corporate credit, market risk and credit instruments, retail/consumer lending, structured finance, liquidity risk, and others. Examples of key market risk and credit instruments include term structure models, models of stock returns and their volatility, CDS spreads by sector and rating category, etc.

hand, VARs and similar regression-based models are useful for generat-ing a large number

of Monte Carlo simulations. Some regula-tors use VAR for this purpose and choose one of the generated extreme scenarios for stress-testing exercises.

The second class of models is built on microfoundations that are used to arrive at macroeconomic implications of various shocks to an economy. Real business cycle and dynamic stochastic general equilibrium models belong to this class. Central banks around the world often use DSGE models. They are fairly complex and difficult to maintain. Their output is frequently driven by statistical assumptions. Although they explicitly model links in an economy, in-teractions among variables may become difficult to follow and they ultimately may appear as a black box to users.

The Moody’s Analytics model of choice is therefore a system of simultaneous equa-tions, which is a compromise between purely data-driven and purely theory-driven models. It is a structural model in which macroeconomic relationships are captured by regression equations.5 Most variables are forecast in real terms, then deflators are used to produce nominal variables. The level of economic activity is determined by aggregate demand and supply. Short-run

5 A more detailed description of the model is available in a Moody’s Analytics working paper, “Moody’s Analytics Global National Forecast With Alternative Scenarios” by Sunayana Mehra, October 2012.

changes arise because of shifts in aggregate demand, and long-run changes occur when aggregate supply shifts. Finally, potential GDP anchors an economy’s growth trajec-tory (see Chart 1). Population and potential GDP determine the long-term trends, and global prices and GDP are exogenous vari-ables determined outside of the model. We will refer to the main modules as we explain the procedure used to generate forecasts for alternative scenarios.

Interest rates—explicit assumptions for scenarios

Interest rates are key variables in macro-economic alternative scenarios, as they are among the main drivers of economic activ-ity. They determine the level of investment in an economy and can be partly influenced by regulators via monetary policy. For our baseline and standard scenario forecast-ing, we make explicit assumptions about selected interest rates for a given country or region and let our model determine the rest of the rates. The explicit assumptions are translated into a time series for a given rate. This series may be generated using a satel-lite model or it may be driven by assump-tions made for the given country. The series is then exogenized in our model. However, we make use of the model even when we ultimately exogenize a given series.

An example is a policy rate set by a central bank. We first estimate this rate endogenously and then alter it to reflect our expectations of actions by the central bank. In custom and regulatory scenarios, interest rates are often provided by a client

Table 1: U.K. PRA 2013H2 SeriesList of variables for which forecast assumptions are provided by PRA

U.K. real GDPU.K. nominal GDPU.K. consumer pricesU.K. ILO unemployment rateU.K. house pricesU.K. commercial real estate pricesU.K. equity pricesU.K. Bank of England policy rateU.K. 3-mo nominal Treasury Bill yieldU.K. 10-yr nominal Treasury Note yieldU.K. 10-yr real Treasury Note yieldU.K. 3-mo interbank ratesCredit spreads€/£ exchange rate

Sources: PRA, Moody’s Analytcis

11

Chart 1: Moody’s Analytics Country Model Design

Exchange rates

Investment

Wages and salaries

PopulationPrices

GDP

Monetary policy rate

Imports

Government

Exports

Global GDP

Unemployment rate

Consumption

Labor force

Employment

Potential GDP

Other deflators

Import prices

10-yr yield

Global prices

Source: Moody’s Analytics

MOODY’S ANALYTICS

4 February 2014

or a regulator to characterize the nature of a given scenario. Typically, the set of specified rates includes the central bank policy rate, a short-term money market rate, and the 10-year government bond yield. In this case, the objective is to generate forecasts conditional on the given series. The model is used to produce a macroeconomic series consistent with paths of the interest rates.

The underlying data-generating stochas-tic process for interest rates is close to being

nonstationary, which makes their forecasting difficult because estimators of equation coef-ficients do not follow standard distributions in this case 6. This means that usual estima-tion methods such as ordinary least squares may have to be replaced with a more sophis-ticated econo-metric estima-tion technique. However, funda-mentals can be used to explain at least some variation in the

interest rates. We illus-trate their use on two variables, the yield for U.K. Treasury gilts with a maturity of 10 years and the monetary policy rate set by the BoE. The 10-year bond yield is modelled as a function of nominal GDP, the policy rate, and the ratio of budget balance to nominal GDP. We adopt a ver-sion of the so-called

Taylor rule for the policy rate. According to the rule, the central bank sets the policy rate with two objectives in mind: low inflation and low unemployment. These objectives may be contradictory at times, as higher interest rates tend to lower inflation but impede economic performance. Table 2 reports coefficient esti-

6 Technically, a series is nonstationary if it contains a unit root, i.e. a=1 in the data-generating process

mates of the two equations, and Chart 2 dem-onstrates a reasonable historical fit.

As indicated above, though our models can be used to determine interest rate forecasts based on fundamentals, the interest rate path is frequently a part of the regulatory scenario assumptions. In the case of the PRA 2013H2 scenario, PRA provided policy rates for the U.K. and the U.S. for five years.7 We extended this scenario until the end of 2023 and made assumptions regarding the policy rate set by the European Central Bank. All these series were exogenized in our models. The Federal Reserve did not provide the federal funds rate or any other policy rate set by central banks around the world, but it did provide the U.S. three-month Treasury yield. This gives infor-mation used to reverse engineer the policy rate likely to be set by the U.S. central bank. We constructed the policy rate for the U.S. and extended this forecast to other central banks globally, mainly the BoE and the ECB.

Long-term yields for government securi-ties have become increasingly important in light of the European sovereign debt crisis, because they have become a gauge of mac-roeconomic stability. As a result of the crisis, long-term bond yields have frequently devi-ated from fundamentals because of investor sentiment. For example, the cost of borrowing in countries such as Italy has tended to be higher, while the interest rates on German government bonds were lower, as investors viewed Germany as a safe haven during the debt crisis. Therefore, the regulators have typically explicitly set the interest rate path in recent years. The low-interest path in the PRA Anchor 2013H2 exercise is depicted in Chart 3. PRA numbers for the U.S. and the U.K. were expanded to include countries in the euro area and the euro zone aggregate. The policy rates and 10-year yields are used to estimate other interest rates that are included in our models, including short-term money market rates and a number of lending rates.

Exchange ratesSimilar to interest rates, the time series

for exchange rates are fairly close to unit

7 Note that the PRA2013H2 scenario is shifted by six months in this document so that it starts in 2014Q1.

22

0

1

2

3

4

5

6

7

8

9

96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13

Actual 10-yr bond yieldFitted 10-yr bond yieldActual policy rateFitted policy rate

Chart 2: Policy Rates and Yields Fitted vs. ActualU.K. rates, %

Source: Moody’s Analytics

and let our model determine the rest of the rates. The explicit assumptions are translated into a time series for a given rate. This series may be generated using a satellite model or it may be driven by assumptions made for the given country. The series is then exogenized in our model. However, we make use of the model even when we ultimately exogenize a given series.

An example is a policy rate set by a cen-tral bank. We first estimate this rate endoge-nously and then alter it to reflect our expec-tations of actions by the central bank. In cus-tom and regulatory scenarios, interest rates are often provided by a client or a regulator to characterize the nature of a given scenario. Typically, the set of specified rates includes the central bank policy rate, a short-term money market rate, and the 10-year govern-ment bond yield. In this case, the objective is to generate forecasts conditional on the given series. The model is used to produce a mac-roeconomic series consistent with paths of the interest rates.

The underlying data-generating stochastic process for interest rates is close to being nonstationary, which makes their forecasting difficult because estimators of equation coef-ficients do not follow standard distributions in this case 6. This means that usual estimation methods such as ordinary least squares may have to be replaced with a more sophisticat-ed econometric estimation technique. How-ever, fundamentals can be used to explain at least some variation in the interest rates. We illustrate their use on two variables, the yield for U.K. Treasury gilts with a maturity of 10 years and the monetary policy rate set by the BoE. The 10-year bond yield is modelled as a function of nominal GDP, the policy rate, and the ratio of budget balance to nominal GDP. We adopt a version of the so-called Taylor rule for the policy rate. According to the rule, the central bank sets the policy rate with two objectives in mind: low inflation and low un-employment. These objectives may be con-tradictory at times, as higher interest rates tend to lower inflation but impede economic performance. Table 3 reports coefficient es-timates of the two equations, and Chart 2 demonstrates a reasonable historical fit.

As indicated above, though our models can be used to determine interest rate fore-casts based on fundamentals, the interest rate path is frequently a part of the regulato-ry scenario assumptions. In the case of the PRA 2013H2 scenario, PRA provided policy rates for the U.K. and the U.S. for five years.7 We extended this scenario until the end of 2023 and made assumptions regarding the

6 Technically, a series 𝑥𝑥𝑡𝑡 is nonstationary if it contains a unit root, i.e. a=1 in the data-generating process 𝑥𝑥𝑡𝑡 = 𝑎𝑎 𝑥𝑥𝑡𝑡𝑡𝑡 + 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 𝑡𝑡𝑒𝑒𝑒𝑒𝑡𝑡.

7 Note that the PRA2013H2 scenario is shift-ed by six months in this document so that it starts in 2014Q1.

policy rate set by the European Central Bank. All these series were exogenized in our mod-els. The Federal Reserve did not provide the federal funds rate or any other policy rate set by central banks around the world, but it did provide the U.S. three-month Treasury yield. This gives information used to reverse engi-neer the policy rate likely to be set by the U.S. central bank. We constructed the policy rate for the U.S. and extended this forecast to other central banks globally, mainly the BoE and the ECB.

Long-term yields for government securi-ties have become increasingly important in light of the European sovereign debt crisis, because they have become a gauge of mac-roeconomic stability. As a result of the crisis, long-term bond yields have frequently devi-ated from fundamentals because of investor sentiment. For example, the cost of borrow-ing in countries such as Italy has tended to be higher, while the interest rates on German government bonds were lower, as investors viewed Germany as a safe haven during the debt crisis. Therefore, the regulators have typ-ically explicitly set the interest rate path in recent years. The low-interest path in the PRA Anchor 2013H2 exercise is depicted in Chart 3. PRA numbers for the U.S. and the U.K. were expanded to include countries in the euro area and the euro zone aggregate. The policy rates and 10-year yields are used to estimate other interest rates that are included in our models, including short-term money market rates and a number of lending rates.

Exchange rates Similar to interest rates, the time series

for exchange rates are fairly close to unit root series. In the long run, they tend to be driven by growth prospects. In the intermediate term, the main factors affecting exchange rates are inflation and interest rate differen-tials. These factors are embedded in our equation for foreign exchange. Historically, this equation tracks the main trends fairly well, although it underestimates the severity of some short-term fluctuations. Table 4 and Chart 4 show an example equation for the exchange rate of the dollar versus the British pound.

The challenge to forecasting exchange rates lies in their volatile and frequent devia-tion from equilibrium values. Therefore, regu-lators opt to set the key rates by specific as-sumptions, which are a part of regulatory macroeconomic scenarios. In the Compre-hensive Capital Analysis and Review stress-testing exercise, the U.S. Federal Reserve sets the exchange rates of the dollar with respect to the euro, the pound, and the yen (see Chart 5). To expand the set of currencies across all countries, we first calculate implied cross rates.8 Then the rest of the exchange

8 For example, when the rate of the dollar with respect to the pound and the dollar with

rate series is computed using our model equation.

The Fed taper as a simple macroeco-nomic scenario

In principle, macroeconomic scenarios can be constructed by combining assumptions that are interpreted as shocks. Assumptions are translated into time series that are exog-enized. Our model is then re-solved, and we therefore generate forecasts for the remain-ing variables. A construction of scenarios is illustrated by the example of tapering by the Federal Reserve. Since December 2012, the Fed had been buying $85 billion of Treasury and mortgage-backed securities each month. In mid-2013, the Fed hinted that these pur-chases would be reduced and ultimately dis-continued in the near future. Financial mar-kets reacted strongly to this statement, and long-term U.S. Treasury bond yields immedi-ately rose. Interest rates on long-term sover-eign debt for other countries also increased. German yields, however, rose somewhat less than their U.S. counterparts. The interest rate differential is a major short-term driver of the dollar-euro exchange rate. The dollar there-fore immediately appreciated some 3% against the euro and even more against other currencies (See Chart 6).

Financial market volatility quickly eased after the initial flight-to-safety capital flows, but has again reacted strongly in early 2014 as the Fed started cutting its monthly asset purchases. This example illustrates two shocks that can be imposed on the Moody’s Analytics country models. The first shock is an increased interest-rate path and the sec-ond is foreign exchange depreciation against the dollar. We construct a simple scenario that reflects these two shocks but make it more severe to generate a meaningful impact on other variables. We assume that interest rates on the 10-year government bonds of a given country increase by 100 basis points and that the euro depreciates against the dol-lar by 10%. Both changes are permanent. We exogenize the assumptions for long-term yields and the exchange rate and solve our model. The two shocks are propagated by the model equations to the remaining variable. Higher interest rates increase the debt bur-den and reduce government spending, while the euro depreciation boosts exports. Ger-many’s public debt-to-GDP ratio is lower than the euro zone average and relies on ex-ports. Therefore, the combined impact of the two shocks is actually positive. The result is the opposite in Italy, which has high sover-eign debt ratios and relies less on exports (see Chart 7).

Standard, custom and regulatory scenari-os produced by Moody’s Analytics are con-structed in a manner similar to the Fed taper-

respect to euro is given, so is the rate between the pound and euro.

Table 2: Fundamentals Drive Interest Rates in Long Run

Dependent Variable: Real U.K. 10-year yieldR2=0.63Data period: 1995Q4-2013Q3

Coef.Constant 0.72*3-qtr MA (Nominal GDP (y/y)) -0.09Real BoE policy rate 0.84***4-qtr MA(Budget balance ratio to nominal GDP (SA)) - 81.37***

Dependent variable: dlog(Monetary policy rate) R2=0.54Data period: 1998Q2-2013Q3

Constant - 0.09***d(Unemployment gap (SA)) - 0.39***dlog(4-qtr MA (CPI (SA))) 9.42***

Notes: 1. “***” denotes estimate significant at 1%, “**” at 5%, and “*” at 10%, respectively.2. MA stands for Moving Average.3. d stands for a difference of a current variable and its lag. 4. dlog denotes a difference of a natural logarithm.

Source: Moody’s Analytics

MOODY’S ANALYTICS

5 February 2014

root series. In the long run, they tend to be driven by growth prospects. In the inter-mediate term, the main factors affecting exchange rates are inflation and interest rate differentials. These factors are embedded in our equation for foreign exchange. Histori-cally, this equation tracks the main trends fairly well, although it underestimates the severity of some short-term fluctuations. Table 3 and Chart 4 show an example equa-tion for the exchange rate of the dollar versus the British pound.

The challenge to forecasting exchange rates lies in their volatile and frequent de-viation from equilibrium values. Therefore, regulators opt to set the key rates by specific assumptions, which are a part of regulatory macroeconomic scenarios. In the Comprehen-sive Capital Analysis and Review stress-testing exercise, the U.S. Federal Reserve sets the ex-change rates of the dollar with respect to the euro, the pound, and the yen (see Chart 5). To expand the set of currencies across all coun-tries, we first calculate implied cross rates.8 Then the rest of the exchange rate series is computed using our model equation.

The Fed taper as a simple macroeconomic scenario

In principle, macroeconomic scenarios can be constructed by combining assump-tions that are interpreted as shocks. Assump-tions are translated into time series that are exogenized. Our model is then re-solved,

8 For example, when the rate of the dollar with respect to the pound and the dollar with respect to euro is given, so is the rate between the pound and euro.

and we therefore generate forecasts for the remain-ing variables. A construction of scenarios is il-lustrated by the example of taper-ing by the Fed-eral Reserve. Since December 2012, the Fed had been buying $85 billion of Treasury and mortgage-backed securities each month. In mid-2013, the Fed hinted that these purchases would be reduced and ul-timately discontinued in the near future. Financial markets reacted strongly to this statement, and long-term U.S. Treasury bond yields immediately rose. Interest rates on long-term sovereign debt for other coun-tries also increased. German yields, how-ever, rose somewhat less than their U.S. counterparts. The in-terest rate differential is a major short-term driver of the dollar-euro exchange rate. The dollar therefore immediately appreci-ated some 3% against

the euro and even more against other cur-rencies (See Chart 6).

Financial market volatility quickly eased after the initial flight-to-safety capital flows, but has again reacted strongly in early 2014 as the Fed started cutting its

33

0

1

2

3

4

5

6

7

06 08 10 12 14 16F 18F 20F 22F

U.K. U.S. Euro zone Germany Italy

Chart 3: Yields in the Low Path Anchor ScenarioPRA 10-yr government bond yields, %

Sources: ECB, BoE, Federal Reserve, IMF, Bank of Italy, Moody’s Analytics

44

1.4

1.5

1.6

1.7

1.8

1.9

2.0

2.1

96 98 00 02 04 06 08 10 12

Actual $ per £ Fitted $ per £

Chart 4: Good Fit for GBP/USD Exchange Rate$ per £, log(fwtfxiusaq.igbr); R2=0.42, 1998Q1 to 2012Q3

Source: Moody’s Analytics

55

70

80

90

100

110

120

130

08 09 10 11 12 13 14F 15F 16F

£ per $ € per $ ¥ per $

Chart 5: The Fed Sets Main Exchange RatesCCAR exchange rates, Severely Adverse Scenario, 2013Q3=100

Source: Moody’s Analytics

Table 3: Exchange Rate Also Implied by Fundamentals

Dependent variable: log($ per £ exchange rate)R2=0.42Data period: 1998Q1-2012Q3

Coef.Constant - 0.74*** log(U.S.CPI/U.K. CPI) 1.99*** log(U.K. policy rate/U.S. policy rate) - 0.05***

Notes:1. “***” denotes estimate significant at 1%, ”**” at 5%, and “*” at 10%, respectively.2. log denotes a natural logarithm

Source: Moody’s Analytics

MOODY’S ANALYTICS

6 February 2014

monthly asset purchases. This example il-lustrates two shocks that can be imposed on the Moody’s Analytics country models. The first shock is an increased interest-rate path and the second is foreign exchange de-preciation against the dollar. We construct a simple scenario that reflects these two shocks but make it more severe to gener-ate a meaningful impact on other variables. We assume that interest rates on the 10-year government bonds of a given country increase by 100 basis points and that the euro depreciates against the dollar by 10%. Both changes are permanent. We exogenize the assumptions for long-term yields and the exchange rate and solve our model. The two shocks are propagated by the model equations to the remaining variable. Higher interest rates increase the debt burden and reduce government spending, while the euro depreciation boosts exports. Germa-ny’s public debt-to-GDP ratio is lower than the euro zone average and relies on exports. Therefore, the combined impact of the two shocks is actually positive. The result is the opposite in Italy, which has high sovereign debt ratios and relies less on exports (see Chart 7).

Standard, custom and regulatory sce-narios produced by Moody’s Analytics are constructed in a manner similar to the Fed tapering scenario, although the number of shocks capturing the main assumptions is larger. The set of shocks includes all the macroeconomic series given by a regula-tor plus additional assumptions formed by Moody’s Analytics.

Extension of scenarios to other countries and regions

One of the challenges in scenario devel-opment is transmitting a given scenario for one country to others. The expansion process will be described first for the baseline and alternative scenario forecasts. In regulatory scenarios, output series may or may not be provided by the regulator for a particular country or a region. Both possibilities are discussed. When a GDP series is not avail-able, additional analysis of output severity is required and conducted. Finally, we compare various standardized and regulatory scenari-os in the terms of their severity.

In the monthly update of the Moody’s Analytics baseline forecast and six standard scenarios, we first make assumptions about policy rates set by the Federal Reserve, the ECB and the BoE, and exchange rates against the dollar. Further, we generate forecasts for oil prices, potential GDP growth, and other exogenous variables using satellite models. We first use these assumptions to formulate forecasts for the U.S. and then produce fore-casts for other countries once we have all exogenous series available. Our baseline and alternative forecasts are therefore consistent across countries.

In many of the regulatory scenarios, the link between countries is explicitly given. For example, the Fed in its CCAR exercise provides real and nominal GDP growth rates for the U.S. and real GDP growth rates for the U.K., Japan, the euro zone, and develop-ing Asia. A correlation matrix among these series quantifies links among the countries

and regions. Although our standard process does contain linkages among countries, the propagation of shocks is nonlinear, and the ultimate outcome can differ from informa-tion given by the regulator. We therefore use the additional information from the regula-tor to create GDP targets for countries where the output is not supplied by the regulator.

Alternative GDP growth when not specified in a scenario

For example, in the CCAR exercise we simply regressed the growth rate of Ger-man output on its own lag and GDP series provided by the Federal Reserve: the U.S., the U.K., Japan, developing Asia, and the euro zone (see Table 4). The most important driver is euro zone GDP. This simple regres-sion is a model conditional on the series supplied—such a model would not be useful for a standard forecasting exercise, as we would have to also estimate variables on the right-hand side of the equation, either jointly or separately. A similar approach to getting a GDP target was used in 33 countries in the CCAR 2014 exercise.

For euro zone countries we had to make additional adjustments to match the euro area aggregate output given by the Federal Reserve. We use a bottom-up approach for regional aggregates and estimated the aggregate euro area output based on the countries for which we produce forecasts. We calculated the difference between the fore-cast conditional on targets of the euro area countries and the euro zone output series from the U.S. central bank. We divided this

66

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

Jan-08 Oct-08 Sep-09 Jul-10 May-11 Feb-12 Dec-12 Oct-13

U.S. Germany U.K.Fed tapering first mooted in May 2013

Tapering confirmed to start Jan 2014

Chart 6: German Yields Decouple From U.S.’s

Sources: Bloomberg, Moody’s Analytics

10-yr government bond yields, %

77

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Germany France Italy Spain

BaselineBond yield spike 100 bpsEuro 10% depreciationCombined impact

Chart 7: Impact of Fed TaperingReal GDP growth in 2014 under alternative assumptions, %

Source: Moody’s Analytics

MOODY’S ANALYTICS

7 February 2014

residual among the member nations based on a given country’s output and computed an adjusted GDP target.

Output severity with limited available information

For some scenarios such as PRA 2013H2, there is not enough information available to rely solely on the type of regression analysis conducted for the CCAR exercise. In this case, we consider the severity of the U.K. output path given by the PRA.9 Severity is

9 We calculate severity based on simulations from a bivariate vector autoregressive model with a country or region GDP as one of the endogenous series and the world output as the other series. For details see “Estimating Severity of Al-ternative Scenarios” by Petr Zemcik and Ashot Tsharakyan, Regional Financial Review (July 2013).

defined as the probability that contraction or growth of output will be worse than in a given scenario. The challenge is to quantify the severity of a particular narrative in light of historical trends that are seldom as drastic as in the desired scenario. Although it is rath-er straightforward to estimate the likelihood of a recession in any quarter of history, it is only possible to calculate the probability of a projected pattern of growth or decline using numerous alternative forecasts.

Comparing scenarios Table 5 compares the PRA 2013H2 An-

chor scenario and the CCAR 2014 Adverse and Severely Adverse scenarios with the Moody’s Analytics Euro Zone Breakup sce-

nario that features a Greek exit. The output start-to-trough decline for the Anchor sce-nario in the U.K. is 4.8%. The probability that the trough would be even deeper is 1.5%; in other words, there is a 1-in-67 chance that the trough will be deeper than 4.8%. The severity of the PRA Anchor scenario is fairly similar to the Moody’s Analytics S4 for the three regions considered—the U.S., the U.K., and the euro zone. The CCAR Severely Ad-verse scenario is slightly less severe that the PRA Anchor scenario: The Adverse scenario’s output decline is the smallest from the four scenarios. When setting up a scenario profile for a given country, historical correlations and severity of a scenario are both consid-ered. Chart 8 demonstrates how the U.K. and U.S. recessions were expanded to the euro area, Germany, and Spain.

GDP is the sum of its componentsGDP is an identity in our model, endog-

enously determined by the sum of its com-ponents: private consumption, investment, inventories, government purchases, exports and imports. Our model equations for house-hold consumption and fixed investment are depicted in Table 6. Per capita consump-tion depends on an interest rate measure, disposable income, oil prices, house prices, and stock prices. The 10-year bond yield is a proxy for the cost of borrowing, oil is a gauge of inflation expectations, and the last two explanatory variables capture wealth effects that are very strong in the U.K. Real invest-ment depends on interest rates, output and stock prices.

88

-7-6-5-4-3-2-1012345

08 10 12 14F 16F 18F 20F 22F

U.K.U.S.Euro zoneGermanySpain

Chart 8: Expansion of Anchor ScenarioPRA GDP, % change yr ago

Source: Moody’s Analytics

Table 5: Anchor Scenario Is the Most SevereReal GDP

U.S. U.K. Euro ZoneStart-to-trough, %PRA -4.6 -4.8 -5.2CCAR Adverse -0.4 -0.9 -1.6CCAR Severely Adverse -2.1 -3.7 -3.9S4 -4.3 -4.4 -4.8

Probability of a worse start-to-trough, %PRA 5 2 3CCAR Adverse 35 16 23CCAR Severely Adverse 17 3 8S4 6 2 4

Sources: PRA, Federal Reserve, Moody’s Analytics

Table 4: German and Euro Zone Performance Closely Connected

Dependent variable: dlog(German real GDP (2005€ bil, SAAR))R2=0.84Data period: 2000Q2-2013Q2

Coef.Constant -0.0003dlog(U.S. real GDP under Fed CCAR baseline (2009$ bil, SAAR)) -0.015dlog(Euro zone real GDP under Fed CCAR baseline (2005€ bil, SAAR)) 1.312***dlog(U.K. real GDP under Fed CCAR baseline (2009£ bil, SAAR)) -0.249*dlog(Japan real GDP under Fed CCAR baseline (2005¥ bil, SAAR)) 0.130*dlog(Developing Asia real GDP under Fed CCAR baseline (2009$ bil, SAAR)) 0.025dlog(Germany real GDP (2005€ bil, SAAR), lag 1) 0.03

Notes:1. “***” denotes estimate significant at 1%, “**” at 5%, and “*” at 10%, respectively.2. dlog is a simple difference of a natural logarithm.

Source: Moody’s Analytics

MOODY’S ANALYTICS

8 February 2014

In the process for generating our stan-dard baseline forecast, we do not target any specific GDP growth rate. This means we start with GDP components and they deter-mine the output path. On the other hand, the challenge in scenario development is to replicate the target profile for GDP using our country model whenever applicable, for example in scenarios provided by regulators.

Decomposition of a given GDP path into components is clearly not unique, and ad-ditional assumptions are needed. A simple decomposition based on historical averages of weights of output components is too sim-plistic, as output components react differ-ently during a crisis. For example, investment fluctuates more than consumption. One possibility is to make assumptions regarding

individual com-ponents of output and then exogenize them one by one. The remaining residual between the sum of the constructed compo-nents and the target GDP can then be distributed among the components ac-cording to their his-torical weight. Chart 9 shows an example

of the PRA Anchor GDP for the U.K. where the output components were estimated fol-lowing the described process.

Automating the Scenario Generation Process

A less time-consuming approach is to leverage existing standard global macroeco-nomic scenarios that Moody’s Analytics up-dates each month. The Moody’s Economic Scenario Accelerator is a tool that has been developed for this purpose.10 MESA uses four standard alternatives scenarios, S1-S4, as inputs, as well as two “bookend” sce-narios that are intentionally constructed to contain very extreme outcomes. Chart 10 illustrates how the input scenarios are used to generate a PRA Anchor scenario. We use the unemployment rate provided by PRA in July 2013 but modify it to account for al-tered economic conditions. The unemploy-

10 See “Moody’s Economic Scenario Accelerator (MESA),” by Tony Hughes and Kyle Hillman, Moody’s Analytics working paper, December 2013.

99

800

900

1,000

1,100

1,200

1,300

1,400

1,500

1,600

1,700

06 08 10 12 14F 16F 18F 20F 22F

C C+I C+I+G C+I+G+(X-M)=GDP

Chart 9: High Consumption in the U.K.PRA U.K. GDP components, 2009£ bil, SAAR

Sources: ONS, Moody’s Analytics

Table 6: Equation Specifications for Investment and Consumption in the U.K. Model

Dependent variable: dlog(Real household consumption per capita)R2=0.56Data period: 1995Q3-2012Q2

Coef.3-qtr MA(d(Real 10-yr bond yield (NSA))) -0.003pdl(dlog(Real disposable income per capita (2005£ bil, SA),3) 0.115***dlog(4-qtr MA of Brent crude oil futures price ($ per barrel, NSA), lag 2) -0.006dlog(Average nominal house prices (£, SA)) 0.137***dlog(FTSE 100 Index) 0.028**

Dependent variable: dlog(Real fixed investment) R2=0.36, 1999Q2-2012Q2R2=0.36Data period: 1999Q2-2012Q2

d(1-qtr MA of Real BoE discount rate (NSA)) 0.008dlog(Real GDP (2009£ bil, SAAR)), lag 1 1.155***dlog(FTSE 100 Index), lag 2 0.115***

Notes1. “***” denotes estimate significant at 1%, “**” at 5%, and “*” at 10%, respectively.2. MA denotes Moving Average.3. log stands for natural logarithm.4. dlog stands for the simple difference of a natural logarithm.5. pdl (…,#) stands for a polynomial distributed lag of order #.6. d is a simple difference of a variable and its own lag.

Source: Moody’s Analytics

MOODY’S ANALYTICS

9 February 2014

ment rate in 2020Q4 is 8.6% in the Anchor scenario and can be expressed as a linear combination of unemployment rates in S4 and in the extreme bookend (MAX) scenario at the same time (see Chart 10). Similar combinations exist at all points in time, and the unemployment rate can cross the input scenario series.

The same linear combination at a partic-ular point in time is then applied to all se-ries that can be exogenized in the Moody’s Analytics country model. The model is re-solved to forecast the remaining variables. This generates a realistic scenario that is passed on to the country analyst. Although MESA is not an economic model in a tradi-tional sense, it draws on properties of exist-ing scenarios that are internally consistent and not overlapping. As a result, a scenario generated by MESA is approximately inter-nally consistent as well. As the GDP series cannot be exogenized, the country analyst

will still need to redistribute the residual between the country model’s GDP and the target GDP among the GDP components, but the residual is very small in this case. MESA was used on a large-scale basis in the CCAR exercise. Chart 11 illustrates how the GDP paths for the U.S., the U.K., and the euro zone—all of which were provided by the Federal Reserve—compare with the target GDP for Germany and Spain, which were calculated by Moody’s Analytics.

Extension to other variables: Beyond the core

The core model variables are used to project other variables such as house prices. The equation for the process driving resi-dential real estate prices is similar across countries, although details of the specifi-cation differ based on data availability. In most places, including the U.K., house prices depend on mortgage rates or their proxy,

personal income, and unemployment rates. For the Anchor exercise, the model equation is not used for the U.K., as the house price path is specified by the PRA and the series is therefore exogenized. For other coun-tries, the equation is used to forecast house prices, conditional on exogenized series (see Chart 12). On the other hand, the model is employed to calculate retail sales or other variables of interest, for the regulator does not provide a particular target series. The un-employment rate depends on the difference between potential and actual GDP. However, we also calculate the unemployment rates for regional aggregates of countries, in which case we use a bottom-up approach by pro-jecting the individual country forecasts and adding up. Thus, the unemployment rate is calculated as the sum of the number of unemployed in countries within the region divided by the sum of the labour force in the region (see Chart 13).

1313

2

4

6

8

10

12

14

16

18

08 09 10 11 12 13 14F 15F 16F

Europe North AmericaEuro zone WorldLatin America Developed AsiaDeveloping Asia

Chart 13: CCAR UnemploymentCCAR unemployment rate, %

Source: Moody’s Analytics

1111

-7-6-5-4-3-2-10123456

08 09 10 11 12 13 14F 15F 16F

U.K. U.S. Euro zone Germany Spain

Chart 11: Severely Adverse ScenarioCCAR Severely Adverse GDP, % change yr ago

Source: Moody’s Analytics

1212

-18

-15

-12

-9

-6

-3

0

3

6

9

12

08 10 12 14F 16F 18F 20F 22F

U.K.

U.S.

Germany

Spain

Chart 12: House Prices CrashPRA house prices, % change yr ago

Source: Moody’s Analytics

1010

0

5

10

15

20

25

10 12 14F 16F 18F 20F 22F

BL S1 S2 S3 S4 MAX MIN PRA anchor

0.6*S4+0.4*MAX

Chart 10: Economic Scenario AcceleratorU.K. unemployment rate, %

Source: Moody’s Analytics

MOODY’S ANALYTICS

10 February 2014

ConclusionThe process and infrastructure for

development of alternative scenarios by Moody’s Analytics have become increas-ingly efficient and comprehensive over the last decade. They rely on high-quality data from original resources that are updated on a monthly basis so that all newly avail-able data are incorporated; forecasts are updated each month. The general template

used for each country model has been revised recently, equation coefficients are re-estimated annually, and the model equa-tions are being continuously improved. The infrastructure for data and model storage enables Moody’s Analytics to provide rapid deliveries to a client. The process leverages highly qualified country analysts in the stage of formulation of assumptions and on the quality of the country models. Gener-

ated forecasts undergo a rigorous quality control check, and satellite models provide additional support. Variables such as inter-est rates and exchange rates are subject to additional consistency checks. The macro-economic forecasts are also often used as inputs into other types of models relevant for clients. The process is thorough but suf-ficiently flexible to produce scenarios with differing available information.

MOODY’S ANALYTICS

11 February 2014

Appendix 1: 2013H2 PRA Anchor Scenario for the U.K.

U.K. scenario summary » A deep U.K. recession is triggered by

contagion from a reintensification of euro area sovereign concerns.

» The escalating financial turmoil in Europe and worsening banking crisis magnify the shortage of liquidity in the U.K. banking system, putting upward pressure on interest rates for consumers and firms in early 2014.

» Short-term funding pressures prompt swift action from the Bank of England, helping to ease the spike in credit spreads by mid-2014.

» Reduced credit availability weighs on consumer spending and fixed investment.

» The sharp buildup in national debt since the 2008-2009 recession prevents the U.K. government from mounting a stimulus in response to combat the deepening recession.

» Real GDP contracts for eight quarters beginning in the first quarter of 2014, cumulatively declining by 5.7%, less than the depth of its decline during the 2008-2009 recession, but still more than in most other recessions.

» The unemployment rate peaks at 11.4% at the end of 2016 and remains in double digits until 2019.

» House prices fall by more than 20% during the first two years of the sce-nario, and the FTSE 100 Index falls by around 25% over the same period and remains volatile thereafter.

TriggerFinancial contagion from the re-intensifi-

cation of euro zone sovereign debt concerns, via risk aversion, is swiftly transmitted to the U.K. The tightening of credit conditions squeezes domestic spending, especially in credit-sensitive markets such as housing, reversing the nascent recovery in Britain’s housing market. The decline in exports to the euro zone also contributes to the U.K. down-turn and has widespread negative impacts via the effect on confidence and uncertainty.

Impact on sovereign yields and foreign exchange rates

Short-term funding pressures prompt swift action from the BoE in early 2014. Despite the central bank acting to provide a substantial monetary stimulus, the reluctance of banks to lend complicates the transmission of ultra- loose monetary policy into greater invest-ment and economic activity. The previous buildup in national debt following the 2008-2009 recession—as Britain’s national debt-to-GDP ratio ballooned from below 45% in 2008 to around 90% in 2013—prevents policymak-ers from mounting a stimulus package in response. Although the deep recession causes the U.K. government’s deficit and debt ratios to rise, financial markets continue to view the U.K. as a safe haven. Also, the Bank of England aggressively expands its asset-purchasing program, providing further support for low bond yields, limiting the upward pressure on government borrowing costs and supporting the British pound.

Impact on banking sectorThe escalation of the euro zone sovereign

debt crisis puts pressure on the U.K. banking system. More than half of U.K. bank assets are in Europe. The large exposure of U.K. banks to the indebted euro zone nations sends shock waves through Britain’s financial sector. Investor uncertainty and Britain’s close financial links to the euro area create severe problems in the country’s interbank market and liquidity shortfalls for many banks. Banks’ asset quality also deteriorates because of losses on cross-border exposures. Rising nonperforming loans at home and abroad put pressure on capital adequacy ra-tios, fueling the banking crisis in the U.K.

Impact on economy and asset pricesThe deepening European and global re-

cession reduces U.K. exports and therefore industrial production. As a result, employment declines sharply, reducing consumer sentiment and overall consumer spending as households return to retrenchment. Real GDP contracts for eight straight quarters from the first quarter

of 2014, posting a 5.7% cumulative drop. The combination of recession, rising unemploy-ment, and reduced credit availability causes residential investment to fall and fuels another wave of house price declines. In this scenario, U.K. house prices, as measured by Halifax’s av-erage house price, fall by around 27%, result-ing in a decline of 41% from their historic peak in the third quarter of 2007. Britain’s stock market also suffers large losses in this scenario because of a sharp slide in nonresidential in-vestment and corporate profits, with the FTSE 100 index falling by around 25%— somewhat shallower than the 33% decline reported dur-ing the 2008-2009 financial crisis—but it re-mains volatile for several years. The unravelling of asset prices leads to a renewed surge in non-performing loans and further destabilises the U.K. financial sector. Consumer prices remain essentially flat for several years amid stagnant demand, but weak productivity growth in the U.K. ensures that the economy does not expe-rience deflation. The BoE keeps the monetary policy rate on hold at 0.5% until the third quarter of 2019, when it starts rising after GDP begins to grow at a steady pace slightly below the potential growth rate.

RecoveryThe deep contraction in U.K. output is fol-

lowed by a slow recovery. Britain’s economy begins to grow again in 2016 as the global economy starts to recover and domestic busi-ness confidence bottoms out, but growth remains well below potential until 2019. The prolonged slump in the U.K. housing market, weak labour market and income growth, and the lack of a fiscal stimulus result in an unusu-ally slow recovery. It will take several years for domestic demand to start to pick up, given the strong fall in property prices, renewed credit tightening, and lingering labour market weakness. The slow recoveries of European trading partners further restrain the U.K. economic recovery. As the recovery is very weak for several years, the unemployment rate remains in the double digits until 2019. Economic growth eventually returns close to trend towards the end of the scenario.

MOODY’S ANALYTICS

12 February 2014

Appendix 2: Selected Variables in the U.K. ModelTrade: Nominal exchange rate (€ per £, NSA)Trade: Nominal exchange rate (¥ per £, NSA)Trade: Nominal exchange rate ($ per £, NSA)National accounts: Real private consumption (2009£ bil, SAAR)National accounts: Real government consumption (2009£ bil, SAAR)National accounts: Real total exports (2009£ bil, SAAR)National accounts: Real total imports (2009£ bil, SAAR)National accounts: Real net exports (2009£ bil, SAAR)National accounts: Real gross domestic product at market prices (2009£ bil, SAAR)National accounts: Fixed investment (2009£ bil, SAAR)National accounts: Change in inventories (2009£ bil, SAAR)National accounts: Real gross capital formation (2009£ bil, SAAR)National accounts: Real domestic demand (2009£ bil, SAAR)National accounts: Nominal private consumption (£ bil, SAAR)National accounts: Nominal government consumption (£ bil, SAAR)National accounts: Fixed investment (£ bil, SAAR)National accounts: Change in inventories (£ bil, SAAR)National accounts: Nominal gross capital formation (£ bil, SAAR)National accounts: Nominal total exports (£ bil, SAAR)National accounts: Nominal total imports (£ bil, SAAR)National accounts: Nominal net exports (£ bil, SAAR)National accounts: Nominal gross domestic product at market prices (£ bil, SAAR)National accounts: Nominal domestic demand (£ bil, SAAR)Implicit price deflator: Private consumption (2009=100, SA)Implicit price deflator: Government consumption (2009=100, SA)Implicit price deflator: Gross capital formation (2009=100, SA)Implicit price deflator: Total exports (2009=100, SA)Implicit price deflator: Total imports (2009=100, SA)Implicit price deflator: Gross domestic product at market prices (2009=100, SA)National accounts: Real private consumption (2009$ bil, SAAR)National accounts: Real government consumption (2009$ bil, SAAR)National accounts: Real gross capital formation (2009$ bil, SAAR)National accounts: Real gross capital formation (2009$ bil, SAAR)National accounts: Real total exports (2009$ bil, SAAR)National accounts: Real total imports (2009$ bil, SAAR)National accounts: Real gross domestic product at market prices (2009$ bil, SAAR)National accounts: Real net exports (2009$ bil, SAAR)National accounts: Real domestic demand (2009$ bil, SAAR)National accounts: Nominal private consumption ($ bil, SAAR)National accounts: Nominal government consumption ($ bil, SAAR)National accounts: Nominal total exports ($ bil, SAAR)National accounts: Nominal total imports ($ bil, SAAR)National accounts: Nominal gross domestic product at market prices ($ bil, SAAR)National accounts: Nominal net exports ($ bil, SAAR)National accounts: Nominal gross capital formation ($ bil, SAAR)National accounts: Nominal domestic demand ($ bil, SAAR)National accounts: Real GDP discrepancy (2009£ bil, SAAR)Consumer price index - Harmonized (2005=100, SA)Household unsecured debt: Nominal credit card lending - amount

outstanding (£ bil, SA)Debt service ratio - Nominal credit card debt to nominal disposable

income (%, SA)Household debt: Nominal lending secured by dwellings (£ bil, SA)Debt service ratio - Nominal mortgage debt to nominal disposable income (%, SA)Household unsecured debt: Other nominal amount outstanding (£ bil, SA)Debt service ratio - Other nominal consumer debt to nominal disposable

income (%, SA)

Household debt: Monthly amounts outstanding of total sterling net lending to individuals and housing associations (£ bil, SA)

Total debt service ratio (%, SA)Household unsecured debt: Monthly amounts outstanding of total sterling net

unsecured lending to individuals, £ mil, SALabour force survey: EUROSTAT unemployment rate (%, SA)Government finance: Budget balance (£ bil, SA)Government finance: Debt outstanding (£ bil, SA)Government finance: Expenditure - total (£ bil, SA)Government finance: Interest payments (£ bil, SA)Government finance: Revenue - total (£ bil, SA)Housing: Nominal mortgage refinancings (£ bil, SA)Housing: Average nominal house prices (£, SA)Housing: Starts in England only (ths, SA)Industrial production: Total (2009=100, SA)Labour force survey: Total employment (mil, SA)Labour force survey: Labour force (mil, SA)Labour force survey: Unemployment rate (%, SA)Labour force survey: Total unemployed (mil, SA)Money: M4 (£ bil, SA)Housing: Nominal mortgage equity withdrawal (£ bil, SA)Producer price index: Input prices (materials and fuel) (2000=100, SA)World development indicators: Purchasing power parity - PPP conversion

factor to official exchange rate ratio (ratio)Interest rate: Credit card loans to households (%, NSA)Interest rate: 10-yr discount bond yield (%, NSA)Interest rates: Libor on £ - 1 yr (%, NSA)Interest rates: Libor on £ - 3 mo (%, NSA)Interest rate: Money market rate (%, NSA)Mortgage rate: 5-yr fixed rate (%, NSA)Mortgage rate: Weighted effective average rate (%, NSA)Mortgage rate: Variable rate (%, NSA)Interest rate: Official discount rate - Bank of England (%, NSA)Interest rate: Other loans to households (%, NSA)Retail sales index: All retailers (2008=100, SA)Stock market: FTSE 100 IndexBOP: Nominal current account balance ($ bil, SA)BOP: Nominal current account balance (£ bil, SA)Personal income: Nominal disposable income (£ bil, SA)Personal income: Wages and salaries (2009$ bil, SA)Personal income: Wages and salaries ($ bil, SA)Personal income: Real wages and salaries (2008£ bil, SA)Personal income: Nominal wages and salaries (£ bil, SA)Interest rates: Loans to nonfinancial corporations - weighted average (%, NSA)Personal income: Real disposable income (2008£ bil, SA)National accounts: Gross domestic product (International $ bil, SAAR)National accounts: Gross domestic product (2000 international $ bil, SAAR)Business investment: Gross fixed capital formation (£ bil, CVM, SA)Unemployment-related benefits: Claimant count rate - All claimants (%, SA) Labour market statistics: Integrated FR - average weekly earnings - total pay (£,

SA)Retail price index: All items (Jan 1987=100, NSA)Retail sales: Value - Total ex sale; maintenance and repair of motor vehicles and

motorcycles (£ bil, SAAR)Retail sales: Volume in constant 2010 prices - Total ex sale; maintenance and

repair of motor vehicles and motorcycles (2010£ bil, SAAR)Employment - Total (mil, SA)

Source: Moody’s Analytics

MOODY’S ANALYTICS

About the Author

Petr Zemcik supervises a team of economists in the Moody’s Analytics London and Prague offices and is currently responsible for quality control and validation of mac-roeconomic country models and related products and services in Europe. He previously worked at CERGE-EI, a joint workplace of the Center for Economic Research and Graduate Education of Charles University in Prague and the Economics Institute of the Academy of Sciences of the Czech Republic, and at Southern Illinois University in Carbondale. He has published numerous articles on econometric methodology and on real estate bubbles in the United States and in Europe in peer-reviewed professional journals. He holds a PhD and MA in Economics from the University of Pittsburgh and an MSc in Econometrics and Operations Research from the University of Economics in Prague.

MOODY’S ANALYTICS

About Moody’s Analytics

Moody’s Analytics helps capital markets and credit risk management professionals worldwide respond to an evolving

marketplace with confi dence. With its team of economists, the company offers unique tools and best practices for

measuring and managing risk through expertise and experience in credit analysis, economic research, and fi nancial

risk management. By offering leading-edge software and advisory services, as well as the proprietary credit research

produced by Moody’s Investors Service, Moody’s Analytics integrates and customizes its offerings to address specifi c

business challenges.

Concise and timely economic research by Moody’s Analytics supports fi rms and policymakers in strategic planning, product and sales forecasting, credit risk and sensitivity management, and investment research. Our economic research publications provide in-depth analysis of the global economy, including the U.S. and all of its state and metropolitan areas, all European countries and their subnational areas, Asia, and the Americas. We track and forecast economic growth and cover specialized topics such as labor markets, housing, consumer spending and credit, output and income, mortgage activity, demographics, central bank behavior, and prices. We also provide real-time monitoring of macroeconomic indicators and analysis on timely topics such as monetary policy and sovereign risk. Our clients include multinational corporations, governments at all levels, central banks, fi nancial regulators, retailers, mutual funds, fi nancial institutions, utilities, residential and commercial real estate fi rms, insurance companies, and professional investors.

Moody’s Analytics added the economic forecasting fi rm Economy.com to its portfolio in 2005. This unit is based in West Chester PA, a suburb of Philadelphia, with offi ces in London, Prague and Sydney. More information is available at www.economy.com.

Moody’s Analytics is a subsidiary of Moody’s Corporation (NYSE: MCO). Further information is available at www.moodysanalytics.com.

About Moody’s Corporation

Moody’s is an essential component of the global capital markets, providing credit ratings, research, tools and analysis that contribute to transparent and integrated fi nancial markets. Moody’s Corporation (NYSE: MCO) is the parent company of Moody’s Investors Service, which provides credit ratings and research covering debt instruments and securities, and Moody’s Analytics, which encompasses the growing array of Moody’s nonratings businesses, including risk management software for fi nancial institutions, quantitative credit analysis tools, economic research and data services, data and analytical tools for the structured fi nance market, and training and other professional services. The corporation, which reported revenue of $3.5 billion in 2015, employs approximately 10,400 people worldwide and maintains a presence in 36 countries.

© 2016, Moody’s Analytics, Moody’s, and all other names, logos, and icons identifying Moody’s Analytics and/or its products and services are trademarks of Moody’s Analytics, Inc. or its affi liates. Third-party trademarks referenced herein are the property of their respective owners. All rights reserved. ALL INFORMATION CONTAINED HEREIN IS PROTECTED BY COPYRIGHT LAW AND NONE OF SUCH INFORMATION MAY BE COPIED OR OTHERWISE REPRODUCED, REPACKAGED, FURTHER TRANSMITTED, TRANSFERRED, DISSEMINATED, REDISTRIBUTED OR RESOLD, OR STORED FOR SUBSEQUENT USE FOR ANY PURPOSE, IN WHOLE OR IN PART, IN ANY FORM OR MANNER OR BY ANY MEANS WHATSOEVER, BY ANY PERSON WITHOUT MOODY’S PRIOR WRITTEN CONSENT. All information contained herein is obtained by Moody’s from sources believed by it to be accurate and reliable. Because of the possibility of human and mechanical error as well as other factors, however, all information contained herein is provided “AS IS” without warranty of any kind. Under no circumstances shall Moody’s have any liability to any person or entity for (a) any loss or damage in whole or in part caused by, resulting from, or relating to, any error (negligent or otherwise) or other circumstance or contingency within or outside the control of Moody’s or any of its directors, offi cers, employees or agents in connection with the procurement, collection, compilation, analysis, interpretation, communication, publication or delivery of any such information, or (b) any direct, indirect, special, consequential, compensatory or incidental damages whatsoever (including without limitation, lost profi ts), even if Moody’s is advised in advance of the possibility of such damages, resulting from the use of or inability to use, any such information. The fi nancial reporting, analysis, projections, observations, and other information contained herein are, and must be construed solely as, statements of opinion and not statements of fact or recommendations to purchase, sell, or hold any securities. NO WARRANTY, EXPRESS OR IMPLIED, AS TO THE ACCURACY, TIMELINESS, COMPLETENESS, MERCHANTABILITY OR FITNESS FOR ANY PARTICULAR PURPOSE OF ANY SUCH OPINION OR INFORMATION IS GIVEN OR MADE BY MOODY’S IN ANY FORM OR MANNER WHATSOEVER. Each opinion must be weighed solely as one factor in any investment decision made by or on behalf of any user of the information contained herein, and each such user must accordingly make its own study and evaluation prior to investing.