Important disclosures and certifications are contained from page 17 of this report. www.danskeresearch.com
Investment Research — General Market Conditions
We introduce the Danske G10 Medium-Term Valuation (MEVA) model which
is a quantitative framework for bridging the short and long term of FX
economics by providing ‘valuation anchors’ for major currency pairs.
MEVA models real bilateral exchange rates based on a selected set of
macroeconomic factors but results are converted to nominal ‘fundamental’
values for relevant currency pairs at a given point in time. We argue that these
are as relevant as directional indicators on a 12-36M horizon.
The G10 MEVA currently points to EUR/USD being undervalued with a model
estimate of around 1.28, hinting that a correction higher in the cross is looming.
In contrast, the moves higher in EUR/SEK and EUR/NOK alike in recent years
are largely warranted by model fundamentals, suggesting that no large
corrections are due.
Given that the MEVA has a half life of around two years, we suggest using the
model signals as a key input for 1-3Y forecasts and associated hedging
recommendations. Specifically, on the basis of, among other things, current
MEVA signals, we recommend EUR-based clients lock in USD income/assets
beyond the 6M horizon. We will publish updates of MEVA on a quarterly basis.
Bridging the gap between the short and the long term
Economic theory provides a range of suggestions as to what determines exchange rates on
different horizons, but with no over-arching theory in place, it remains an empirical
question of what drives FX markets in the short, medium and long term, respectively.
Bridging the gap: FX economics and Danske FX models
Source: Danske Bank Markets
1 October 2015
Senior Analyst, PhD Christin Tuxen +45 45 13 78 67 [email protected]
FX Edge
Introducing the Danske G10 MEVA model
Danske Bank FX models compared
Source: Reviews, DBM. Note: STFM=Danske
Short-Term Financial Model, MEVA= Danske
Medium-Term Valuation Model, PPP=Danske PPP
calculation.
STFM MEVA PPP
EUR/USD -2% -14% -10%
EUR/NOK 2% 9% 13%
EUR/SEK 0% 3% 15%
USD/JPY 2% 13% 31%
GBP/USD -2% -14% -5%
EUR/CHF 0% -15% -21%
AUD/USD 1% -7% 0%
NZD/USD 0% -9% 3%
USD/CAD 3% 13% 9%
Model deviation (% rel to spot)
2 | 1 October 2015 www.danskeresearch.com
FX E
dge
FX Edge
Until now, the Danske FX directional model framework has primarily consisted of two
types of indicators.
First, the Danske Purchasing Power Parity (PPP) calculation offers a long-term fix point
for the currency market. This set-up is based on effective exchange rates which ensures
internally consistent PPP levels among the G10 currencies. Empirically it is well known
that ‘the law of one price holds only in the very long term with half-lives generally found
to at least three-five years.
Second, the Danske FX Short-Term Financial Models (STFM) provide hints of short-run
mispricing. These take into account a range of financial factors including relative interest
rates and are based on 200D rolling estimations. The STFM may be used to identify
misalignments that should be corrected in the near term, say within weeks, see Danske
Bank's FX Quant Strategy 8 April 2015, for the latest update.
Danske PPP calculation points to 1.24
as long-term fix point Danske STFM suggests 1.12 is ‘fair’
Source: Macrobond Financial, DBM Source: Macrobond Financial, DBM
However, the gap between the horizons at which these two types of models may be
employed is potentially rather wide, and neither takes account of real (macroeconomic)
factors which we believe should be key drivers of real exchange rates in the medium
term, i.e. explain deviations from PPP. In this context, the Danske FX Medium-Term
Model (MEVA) framework is meant to be a key input, but not a forecast, for where a
currency pair is likely to be headed on a 12-36M horizon before PPP forces would be
expected to set in. Indeed, at times it does not make sense to incorporate a firm view on
central bank policy divergence (capturing nominal trends) beyond the 12M horizon,
which in turn highlights the need for a fundamental-based anchor (capturing real trends)
in the FX market.
In some cases we may have a clear call on where fundamental factors are headed, for
example, terms of trade, which may imply that we expect the ‘fundamental’ value to
move in a certain direction; this would of course have an impact on our medium-term
view on a cross. But in the absence of such calls and external shocks, the MEVA
framework provides a disciplined way of seeing through the impetus of short-term
factors. We stress that models should be regarded as only one input for forming views and
forecasts, and should not mechanically be employed for making projections.
The paper is organised as follows: we first discuss the MEVA set-up in short followed by
an overview of current model signals and, finally, discuss how to use the model signals.
Details on the estimation procedure and a presentation of the data are left for Appendix A
and B, respectively.
3 | 1 October 2015 www.danskeresearch.com
FX E
dge
FX Edge
Introducing the MEVA - what we do and why in short
The MEVA framework falls within the class of Behavioural Equilibrium Exchange Rate
(BEER) models which estimate equilibrium exchange rates from a set of explanatory
variables. This contrasts with Fundamental Equilibrium Exchange Rate (FEER) models
which calculate equilibrium exchange rates based on assumptions regarding how to
obtain external balance. Our estimation procedure is inspired by the BEER models
developed by the IMF, see inter alia IMF WP/98/67, WP/07/296, WP/08/13, but we make
a crucial amendment to the prevailing IMF set-up:
We following the existing literature in modelling real (rather than nominal) exchange
rates because these are what economic theory has something to say about in the medium
term. But we crucially consider bilateral rather than effective exchange rates as these are
what clients (and we) are ultimately interested in. This also allows for more clarity in
understanding what is driving the ‘fundamental’ model estimates.
Importantly, we take a panel (rather than a single-equation) approach to estimation, see
Appendix A for details: this effectively entails that we assume that the dynamics of G10
currencies can be described by one model, i.e. that the relationship between the real
exchange rate and medium-term fundamentals is the same for, say, EUR/USD as it is for,
say, USD/JPY. This assumption may of course be questioned but if (at least
approximately) valid allows for an highly improved model fit.
We build upon IMF work for generating a list of candidate variables to enter the medium-
term equilibrium relationship with the real exchange rate. The G10 MEVA is however
based upon a mere two medium-term drivers: a gauge of the Balassa-Samuelson effect
(see box), and terms of trade. While both of these may be measured by a range of
different series, we use the CPI-PPI differential for the former and the export-import
deflator differential for the latter. When estimated within our model framework, both
factors are found to have the economically right signs (both positive) and be statistically
significant.
EUR/USD as an example: Balassa
Samuelson effect (CPI-PPI)
EUR/USD as an example: terms of
trade
Source: Macrobond Financial, Danske Bank
Markets
Source: Macrobond Financial, Danske Bank
Markets
A few factors not included in the model also deserve mentioning. First, we do not include
a direct measure of external wealth/debt: while this is usually considered a candidate
medium-term factor, our preferred measure for this, the international investment position
(IIP) to GDP, turns out not to be both economically and statistically identified. This may
either be because IIP is an inappropriate measure of external wealth/debt and/or because
another factor such as terms of trade already captures the key dynamics. Second, we do
not include the real interest-rate differential: the rationale is that this captures trends in
relative monetary policy that we arguably want to see through on a medium-term horizon.
Balassa-Samuelson effect explained
Source: Danske Bank Markets
4 | 1 October 2015 www.danskeresearch.com
FX E
dge
FX Edge
For guidance on what relative interest rates and other short-term factors suggest as fair for
the FX market at a given point in time we refer to our STFM framework.
We use monthly data (interpolated if required) from 1985 and onwards albeit for many
countries data series do not start until 1996. The panel consists of AUD/USD, NZD/USD,
USD/CAD, USD/JPY, GBP/USD, EUR/USD, EUR/SEK, EUR/NOK and EUR/CHF.
Note that EUR/DKK is excluded from the estimation due to the Danish peg but that
EUR/CHF is included despite the SNB floor 2011-15. In choosing candidate factors we
emphasise timeliness of data releases, and for data sources we emphasise consistency
across countries/regions. The data is discussed in greater details in Appendix B.
Estimation of the MEVA model broadly takes place in two steps – details are outlined in
Appendix A:
1. We first estimate a static medium-term equilibrium relation between the level of the
real exchange rate and a set of relevant determinants; this provides us with
‘fundamentally justified’ levels for each currency pair which is then converted from
real to nominal terms using the relevant CPI differential.
2. Next, we estimate a dynamic short-term adjustment relation for the change in the
real exchange rate to validate that the real exchange rate adjusts as appropriate in
order to warrant the use of the model as a directional indicator in the medium term.
Separately we estimate the half-life of model deviations.
We note that significant equilibrium correction takes place, thus validating the use of
MEVA estimates as directional indicators. Moreover, we find the half-life of model
deviations to be around two years, suggesting the model is relevant as an anchor on a 12-
36M horizon.
In presenting the model results we report a so-called ‘confidence region’ alongside the
actual spot and MEVA estimates: this is based on +/- 2 standard deviations of the
historical model deviations but should be interpreted with care as the distribution of the
panel model estimator is not known.
Broad overview of the MEVA idea
Source: DBM
5 | 1 October 2015 www.danskeresearch.com
FX E
dge
FX Edge
G10 model results – what we find and what it means
In the following we present an overview of the current model signals for each currency
pair. An overview of model deviations in real versus nominal terms in percentage terms is
given below: in nominal terms, EUR/USD, GBPY/USD and EUR/CHF stand out as the
most undervalued crosses whereas USD/JPY and USD/CAD are markedly overvalued
according to the MEVA framework.
MEVA model deviations in real vs nominal terms
Source: Eviews, DBM
Correction higher is looming for EUR/USD
The model currently suggests that EUR/USD is substantially undervalued with the panel
model suggesting 1.28 as ‘fundamentally justified’. In fact, the ‘fundamental’ value for
EUR/USD has been held up over the past decade by the Balassa-Samuelson effect being
USD-negative (relative to EUR). As oil prices rose in the 2000s, terms of trade moved in
favour of EUR relative to USD but the recent oil-price drop and reduced reliance on
energy imports has made this factor a USD positive over the past year. While the panel
(preferred) and single-equation model for EUR/USD differ slightly (the latter suggests a
‘fundamental’ value of 1.22; not shown here), both hint that a correction higher in the
cross is looming.
-20%-15%-10%
-5%0%5%
10%15%20%
MEVA model deviation in real terms (% rel to spot) in nominal terms
overvalued
undervalued
EUR/USD fundamentally undervalued on MEVA estimates
Source: Eviews, Macrobond Financial, DBM
6 | 1 October 2015 www.danskeresearch.com
FX E
dge
FX Edge
Currently, we expect EUR/USD to head lower short term on monetary-policy divergence
with the Fed set to hike at a time (likely in December) when the ECB may deliver an
extension of the QE scheme. However, history shows that a first Fed hike is not in itself
enough to steer a clear direction for the USD. In this respect, the MEVA estimates
provide a key input to our longer-term forecasts: this is one reason we look for a move
higher in the cross on a 12M horizon, see FX Forecast Update: It ain't over till the Fed
sings.
Fed-ECB divergence set to drag
EUR/USD lower...
... but first Fed hike in itself does not set
a clear direction for USD
Source: Macrobond Financial, DBM Source: Macrobond Financial, DBM
Comparing the signals for EUR/USD from our three different types of models we note
that the MEVA suggests an even higher value for the cross (1.28) than our PPP estimate
(1.25). That is, the MEVA hints that a move above the current PPP level may in fact be
warranted by medium-term factors. The STFM naturally tracks the actual spot rate better
than the other models as we allow for a range of financial factors and a rolling estimation
window, but the caveat is that using financial factors as explanatory variables imply that
we will not be able to spot dislocations in the FX market before a correction takes place
in e.g. the fixed-income market. As such, the STFM, MEVA and PPP estimates should
complement each other well across time horizons.
Scandies undervalued – but not much!
The Scandies have seen model fundamentals move in favour of a higher EUR/NOK and
EUR/SEK alike in recent years. This implies that we can largely explain the up-trends in
the crosses on the basis of a negative Balassa-Samuelson effect and deteriorating terms of
trade. That said, some potential for a move lower in both crosses in the medium term is
nevertheless supported by the MEVA with the ‘fundamental’ value for EUR/NOK and
EUR/SEK at 8.65 and 9.15, respectively.
Fundamentals dragging EUR/NOK
higher in recent years
EUR/SEK uptrend supported by BS
effect and terms of trade since 2013
Source: Eviews, Macrobond Financial, DBM Source: Eviews, Macrobond Financial, DBM
EUR/USD: model inputs at various
horizons
Source: Macrobond Financial, DBM
7 | 1 October 2015 www.danskeresearch.com
FX E
dge
FX Edge
Striking misalignments in other majors
GBP/USD is markedly undervalued currently. Notably the level shift in GBP/USD
following the 2008 financial crisis was not warranted by panel-model drivers with the
Balassa-Samuelson effect in fact GBP-positive in recent years. This suggests that
GBP/USD should move substantially higher towards the 1.70 level in the medium term.
We note however that the panel and single-equation models (not reported here) for cable
differ markedly with the latter suggesting the cross being close to fundamentals.
The actual value of USD/JPY remains within the reported ‘confidence region’ but is
actually at present outside the single-equation model bands (not reported here). Looking
at model fundamentals terms of trade has been a JPY negative for an extended period of
time but the Balassa-Samuelson effect has largely countered this, being a clear positive
since 2008. On balance this leaves USD/JPY markedly overvalued by some 13%
according to MEVA.
The EUR/CHF model results clearly points to the Swissie being overvalued at present,
but we note that results should be interpreted with caution given the change in policy
regime in the period under consideration with the floor on the cross in place 2011-15.
Thus the MEVA hints that EUR/CHF will eventually have to correct higher still with a
move to the mid 1.20s not unreasonable in medium term when/if the deflation issue fades.
GBP/USD undervalued on panel
model... ... while USD/JPY is overvalued
EUR/CHF headed for higher levels -
eventually
Source: Eviews, Macrobond Financial, DBM Source: Eviews, Macrobond Financial, DBM Source: Eviews, Macrobond Financial, DBM
8 | 1 October 2015 www.danskeresearch.com
FX E
dge
FX Edge
Weaker commodity currencies warranted – but CAD looks undervalued
For the major commodity currencies, both the Balassa-Samuelson effect and terms of
trade have been clear positives during the 2000s; this has obviously reversed in recent
years. For both AUD/USD and NZD/USD, the depreciation witnessed over the past few
years is thus largely warranted by fundamentals. However, despite the oil price drop,
USD/CAD is now starting to look somewhat overvalued with MEVA suggesting the
cross is 13% overvalued.
AUD/USD closer to ‘fundamentals’... ... and same goes for NZD/USD USD/CAD model value has risen – but
not as much as spot
Source: Eviews, Macrobond Financial, DBM Source: Eviews, Macrobond Financial, DBM Source: Eviews, Macrobond Financial, DBM
9 | 1 October 2015 www.danskeresearch.com
FX E
dge
FX Edge
FX Strategy: using the MEVA signals – how to
While the MEVA is meant as a tool for bridging short- and the long-term factors, we
stress that the results should be viewed as just one of many inputs in forming longer-term
views on the FX market. With an estimated half-life of around two years, we see the
MEVA framework as a key input for forming directional views on a 12-36M
horizon. Double-digit deviations of current spot values from MEVA estimates are found
in EUR/CHF (15% undervalued), EUR/USD (14% undervalued), GBP/USD (14%
undervalued) as well as in USD/JPY (13% overvalued), USD/CAD (13% overvalued) and
EUR/NOK (10% overvalued).
While CHF will likely remain overvalued for some time still given the largely empty
SNB toolbox, we stress that EUR/USD should be in for a rebound on a 6-12M horizon
provided ECB does not resort to more drastic measures than QE (i.e. stay away from
cutting the deposit rate further). Thus, a key message from the MEVA model at
present for EUR-based clients looking to hedge USD income/assets beyond the 6M
horizon should consider locking in EUR/USD on any drop in relation to a first Fed
hike (likely to arrive in December) as these (low) levels are unlikely to persist longer
term.
Similarly, we stress that while exposed to the pricing of further Norges Bank easing near
term, EUR/NOK is now substantially overvalued on MEVA despite the oil sell-off, and
we maintain that the cross should move lower again further out, see Norges Bank Review:
A 25bp rate cut and an easing bias. Thus while risks are still primarily on the downside
for oil and thus on the upside for EUR/NOK, the pair should not be expected to hover at
these levels for an extended period of time. NOK-based clients with EUR income/assets
should therefore consider locking in current levels of EUR/NOK.
Finally, we note that while USD/JPY is clearly overvalued according to MEVA, we do
not look for an imminent correction of this model deviation as we see reasons why
markets may price continued monetary-policy divergence in favour of USD/JPY for some
time still. Thus, medium-term dynamics as captured by MEVA could take longer to play
out in this case.
Going forward, we shall use the MEVA signals as a medium-term anchor hinting at
where the FX market should be headed in the absence of new shocks to the model
drivers, CPI, or short-term factors such as monetary policy. As such, the MEVA
signals could be used as a guide for longer-term (1-3Y) forecasts and hedging
decisions for corporates and institutions alike. We will publish updates of the MEVA
on a quarterly basis.
Danske FX model estimates compared
Source: Eviews, DBM. Note: STFM=Danske Short-Term Financial Model, MEVA= Danske Medium-Term
Valuation Model, PPP=Danske PPP calculation.
Spot STFM MEVA PPP STFM MEVA PPP
EUR/USD 1.12 1.14 1.28 1.24 -2% -14% -10%
EUR/NOK 9.50 9.27 8.65 8.22 2% 9% 13%
EUR/SEK 9.41 9.42 9.15 7.97 0% 3% 15%
USD/JPY 120.31 118.24 105.00 82.60 2% 13% 31%
GBP/USD 1.52 1.54 1.73 1.60 -2% -14% -5%
EUR/CHF 1.09 1.09 1.26 1.32 0% -15% -21%
AUD/USD 0.70 0.70 0.75 0.70 1% -7% 0%
NZD/USD 0.64 0.64 0.70 0.62 0% -9% 3%
USD/CAD 1.34 1.30 1.16 1.21 3% 13% 9%
Danske model estimates Model deviation (% rel to spot)
10 | 1 October 2015 www.danskeresearch.com
FX E
dge
FX Edge
Appendix A: MEVA estimation procedure – how we do it in detail
In the following we discuss some general modelling considerations and present the
MEVA estimation procedure in details.
First note that we explicitly factor in that PPP does not hold as an equilibrium concept for
the time horizon under consideration, i.e. in the ‘medium term’, which in a statistical
sense imply that the real exchange rate is treated as non-stationary. As a result, we use
cointegration techniques to ensure valid inference.
Regime shifts and data availability limit the historical information on the dynamics
governing each currency pair. Thus, while we may estimate an equilibrium relation for
each currency pair in isolation, we argue that pooling information across G10 currency
pairs while assuming a homogenous equilibrium relation for these, i.e. we take a panel
approach to estimation, arguing that this provides a more informed modelling procedure.
We shall thus use the G10 panel model as the main reference for judging how far off
“fundamental” value a currency pair is at any given point in time, but individual (single-
equation) models are used as a cross-check and as a guide to model uncertainty. Note that
confidence bans are not readily available for the panel cointegration model as the
distribution of the panel estimator is not known.
Also, note that while we estimate the model in real terms we are interested in the nominal
exchange rate. In conducting this real-to-nominal transformation of the model estimates
we use the relevant CPI differential at any given point in time. In doing this, we
essentially impose medium-term homogeneity between the nominal exchange rate and the
CPI differential. Furthermore, when arguing that the MEVA ‘equilibrium error’ tells us
how the nominal exchange rate should correct over time, we are inherently assuming that
adjustment takes place solely via the exchange rate, which may not necessarily be the
case. That said, CPI differentials tend to be rather persistent compared with nominal
exchange rates (i.e. movement in the real exchange rate is primarily caused by the
nominal exchange rate), suggesting this is a reasonable assumption in the medium term.
Finally, the panel model structure makes it possible to calculate a model-implied value for
currency pairs not directly included in estimation of the model. While USD/SEK, for
example, should not be included in estimating the model due to multi-collinearity
(EUR/USD and EUR/SEK included already), the panel structure of the model suggests
that the estimated relations should be more generic in nature, and thus that is may be
justifiable to use these for calculating ‘fundamental’ levels for other crosses. Thus, we
may calculate a model-implied value for USD/SEK based on US versus Swedish
fundamental factors (conditional on certain assumptions about the individual-specific
intercept) or we could simply calculate it from the model-implied values for EUR/USD
and EUR/SEK.
Below, we outline the econometric details of the MEVA estimation procedure.
Real vs nominal EUR/USD
Source: Macrobond Financial, DBM
11 | 1 October 2015 www.danskeresearch.com
FX E
dge
FX Edge
1. Single-equation estimation: a first glance at relations (cross check)
Estimate a single cointegrating relation for each currency pair individually (equation-
by-equation) by Fully Modified OLS (FMOLS with a restricted constant); notably
this provides confidence bands (note that panel estimation does not).
Test for cointegration and equilibrium correction using standard OLS.
2. Panel model estimation: Danske MEVA framework
Test for the number of cointegrating relations and check at most one cointegrating
relation is present (Johansen-Fisher test).
Estimate the panel cointegration model by weighted FMOLS (restricted constant).
Check for economic and statistical identification by judging the validity of the
parameter signs and their significance, respectively.
3. Testing validity of MEVA
Test for cointegration using panel unit root test (Im-Pesaran-Shin test).
Test for equilibrium correction (fixed-effects OLS).
Calculate half-life of equilibrium error (fixed-effects OLS).
4. Using MEVA
Transform model estimates from real to nominal using relevant CPI differentials.
Compare model ('fundamental') values to actual ones; note that confidence bands are
not readily available as the distribution of the estimator is non-standard and requires
simulation; instead we report a ‘confidence region’ based on the standard error of the
equilibrium error.
Use for conditional forecasting (requires forecasts of the explanatory factors).
Overview of full MEVA estimation procedure
Source: DBM
EUR/USD: single-equation vs G10
panel results
Source: Eviews, Macrobond Financial, DBM
12 | 1 October 2015 www.danskeresearch.com
FX E
dge
FX Edge
An overview of the modelling procedure in equations is given below.
Danske G10 MEVA procedure in equations
Source: DBM
13 | 1 October 2015 www.danskeresearch.com
FX E
dge
FX Edge
Appendix B: G10 data series – what are the drivers?
We briefly discuss potential medium-term drivers and then present the data series used in
the G10 MEVA.
Balassa-Samuelson effects. Theoretically, these are caused by productivity
differences between the tradable and non-tradable sector leading to disproportional
upward pressure on wages and prices in the non-tradable sector, thus explaining real
appreciation. This may be captured by the relative CPI-PPI differential (as PPI
includes mainly tradables whereas CPI captures both) or real-income differentials (cf.
Penn effect). Alternatively, differentials in labour productivity or in real income could
be used. Another factor potentially capturing (part of) this effect is the ratio of
government consumption (falling more intensively on non-tradables) to GDP. In the
G10 MEVA we use CPI-PPI differentials.
Terms of trade. Movements in the price of exports relative to imports caused by, for
example, commodity-price moves may have significant wealth effects for a country;
as such terms of trade are related to the accumulation of net foreign assets. In the G10
MEVA we use export-import deflator differentials.
One more candidate factor warrants a brief discussion, namely external
wealth/indebtedness. A country accumulating foreign assets will typically see a stronger
exchange rate and thus run larger current-account deficits, ceteris paribus. This effect may
be captured by various measures of net foreign assets but such data are usually not readily
available and/or frequently updated. The ‘net international investment position’ is an
alternative and more timely indicator available from national sources but this does not
account for valuation changes, capital transfers, debt reductions etc. Finally, an
alternative to these stock measures is the current-account balance (flow). All such
measures should be measured relative to GDP or total trade to make cross-country
comparison reasonable. In the G10 MEVA, the ‘external wealth’ factor is excluded as it is
not properly identified in both an economic and statistical sense simultaneously.
US: net foreign asset vs IIP Eurozone: net foreign asset vs IIP
Source: LMF(2006), Macrobond Financial, DBM Source: LMF(2006), Macrobond Financial, DBM
The data series are graphed below. Note that all series are log-transformed and relatively
defined, i.e. for EUR/USD we consider the euro area versus the US series. Exchange rates
are monthly averages of daily closing prices and converted from nominal to real using
CPI deflators from national sources; PPI, export and import deflators are OECD data.
EUR/USD: real exchange rate vs
relative IIP
Source: Macrobond Financial, DBM
14 | 1 October 2015 www.danskeresearch.com
FX E
dge
FX Edge
US vs. EU: Balassa Samuelson effect
(CPI-PPI) US vs. EU: terms of trade
Source: Macrobond Financial, DBM Source: Macrobond Financial, DBM
Sweden vs. EU: Balassa Samuelson
effect (CPI-PPI) Sweden vs. EU: terms of trade
Source: Macrobond Financial, DBM Source: Macrobond Financial, DBM
Norway vs EU: Balassa Samuelson
effect (CPI-PPI) Norway vs EU: terms of trade
Source: Macrobond Financial, DBM Source: Macrobond Financial, DBM
15 | 1 October 2015 www.danskeresearch.com
FX E
dge
FX Edge
Swiss vs EU: Balassa Samuelson
effect (CPI-PPI) Swiss vs EU: terms of trade
Source: Macrobond Financial, DBM Source: Macrobond Financial, DBM
UK vs US: Balassa Samuelson effect
(CPI-PPI) UK vs US: terms of trade
Source: Macrobond Financial, DBM Source: Macrobond Financial, DBM
Japan vs US: Balassa Samuelson
effect (CPI-PPI) Japan vs US: terms of trade
Source: Macrobond Financial, DBM Source: Macrobond Financial, DBM
16 | 1 October 2015 www.danskeresearch.com
FX E
dge
FX Edge
Australia vs US: Balassa Samuelson
effect (CPI-PPI) Australia vs US: terms of trade
Source: Macrobond Financial, DBM Source: Macrobond Financial, DBM
New Zealand vs US: Balassa
Samuelson effect (CPI-PPI) New Zealand vs US: terms of trade
Source: Macrobond Financial, DBM Source: Macrobond Financial, DBM
Canada vs US: Balassa Samuelson
effect (CPI-PPI) Canada vs US: terms of trade
Source: Macrobond Financial, DBM Source: Macrobond Financial, DBM
17 | 1 October 2015 www.danskeresearch.com
FX E
dge
FX Edge
Disclosures This research report has been prepared by Danske Bank Markets, a division of Danske Bank A/S (‘Danske
Bank’). The author of the research report is Christin Tuxen, Senior Analyst.
Analyst certification
Each research analyst responsible for the content of this research report certifies that the views expressed in the
research report accurately reflect the research analyst’s personal view about the financial instruments and issuers
covered by the research report. Each responsible research analyst further certifies that no part of the compensation
of the research analyst was, is or will be, directly or indirectly, related to the specific recommendations expressed
in the research report.
Regulation
Danske Bank is authorised and subject to regulation by the Danish Financial Supervisory Authority and is subject
to the rules and regulation of the relevant regulators in all other jurisdictions where it conducts business. Danske
Bank is subject to limited regulation by the Financial Conduct Authority and the Prudential Regulation Authority
(UK). Details on the extent of the regulation by the Financial Conduct Authority and the Prudential Regulation
Authority are available from Danske Bank on request.
The research reports of Danske Bank are prepared in accordance with the Danish Society of Financial Analysts’
rules of ethics and the recommendations of the Danish Securities Dealers Association.
Conflicts of interest
Danske Bank has established procedures to prevent conflicts of interest and to ensure the provision of high-
quality research based on research objectivity and independence. These procedures are documented in Danske
Bank’s research policies. Employees within Danske Bank’s Research Departments have been instructed that any
request that might impair the objectivity and independence of research shall be referred to Research Management
and the Compliance Department. Danske Bank’s Research Departments are organised independently from and do
not report to other business areas within Danske Bank.
Research analysts are remunerated in part based on the overall profitability of Danske Bank, which includes
investment banking revenues, but do not receive bonuses or other remuneration linked to specific corporate
finance or debt capital transactions.
Financial models and/or methodology used in this research report
Calculations and presentations in this research report are based on standard econometric tools and methodology
as well as publicly available statistics for each individual security, issuer and/or country. Documentation can be
obtained from the authors on request.
Risk warning
Major risks connected with recommendations or opinions in this research report, including a sensitivity analysis
of relevant assumptions, are stated throughout the text.
Expected updates
This publication is updated on a weekly basis.
Date of first publication
See the front page of this research report for the date of first publication.
General disclaimer This research has been prepared by Danske Bank Markets (a division of Danske Bank A/S). It is provided for
informational purposes only. It does not constitute or form part of, and shall under no circumstances be
considered as, an offer to sell or a solicitation of an offer to purchase or sell any relevant financial instruments
(i.e. financial instruments mentioned herein or other financial instruments of any issuer mentioned herein and/or
options, warrants, rights or other interests with respect to any such financial instruments) (‘Relevant Financial
Instruments’).
The research report has been prepared independently and solely on the basis of publicly available information that
Danske Bank considers to be reliable. While reasonable care has been taken to ensure that its contents are not
untrue or misleading, no representation is made as to its accuracy or completeness and Danske Bank, its affiliates
and subsidiaries accept no liability whatsoever for any direct or consequential loss, including without limitation
any loss of profits, arising from reliance on this research report.
18 | 1 October 2015 www.danskeresearch.com
FX E
dge
FX Edge
The opinions expressed herein are the opinions of the research analysts responsible for the research report and
reflect their judgement as of the date hereof. These opinions are subject to change, and Danske Bank does not
undertake to notify any recipient of this research report of any such change nor of any other changes related to the
information provided in this research report.
This research report is not intended for retail customers in the United Kingdom or the United States.
This research report is protected by copyright and is intended solely for the designated addressee. It may not be
reproduced or distributed, in whole or in part, by any recipient for any purpose without Danske Bank’s prior
written consent.
Disclaimer related to distribution in the United States This research report is distributed in the United States by Danske Markets Inc., a U.S. registered broker-dealer
and subsidiary of Danske Bank, pursuant to SEC Rule 15a-6 and related interpretations issued by the U.S.
Securities and Exchange Commission. The research report is intended for distribution in the United States solely
to ‘U.S. institutional investors’ as defined in SEC Rule 15a-6. Danske Markets Inc. accepts responsibility for this
research report in connection with distribution in the United States solely to ‘U.S. institutional investors’.
Danske Bank is not subject to U.S. rules with regard to the preparation of research reports and the independence
of research analysts. In addition, the research analysts of Danske Bank who have prepared this research report are
not registered or qualified as research analysts with the NYSE or FINRA but satisfy the applicable requirements
of a non-U.S. jurisdiction.
Any U.S. investor recipient of this research report who wishes to purchase or sell any Relevant Financial
Instrument may do so only by contacting Danske Markets Inc. directly and should be aware that investing in non-
U.S. financial instruments may entail certain risks. Financial instruments of non-U.S. issuers may not be
registered with the U.S. Securities and Exchange Commission and may not be subject to the reporting and
auditing standards of the U.S. Securities and Exchange Commission.