Research Methodologies

Latest update: November 14, 2024

Qualitative methodologies

Our economic projections are based on a combination of (country) expert opinion, small-scale models and the NiGEM DSGE (Dynamic Stochastic General Equilibrium) model. Typically, the qualitative part is about choosing the amount of fiscal stimulus/contraction, taking into account the potential impact of certain events (elections) or structural changes (reforms for example). In recent years, we have also developed a geopolitical ‘framework’ that has a qualitative bearing on our forecasts (for example, a modest ‘add-on’ to inflation to reflect potential supply-chain disruptions etc.)

The DSGE model puts most emphasis on the demand side of the economy; since Covid-19 we have added a bottom-up sectoral supply side analysis. This is largely a qualitative analysis (deciding which sectors are mostly affected by containment measures, for example). For short-term forecasting/analysis (US/Europe) we also regularly cross-check with recession-probability (logit/probit) models.

Long-term economic projections are based on productivity trends, labour force and innovation.

Inflation projections are also based on a combination of expert judgement, special factors/events (like VAT adjustments) and a model that links up our projections for commodity prices (key inputs being RaboResearch energy price forecasts), FX rates and economic growth/the output gap.

Regarding money market projections, our projections for key central bank rates are based on a combination of expert judgement, small scale models (Taylor rule framework, a model for money market forward and credit spreads). We combine this framework with analysis of pronouncements by ECB officials (for example whether the emphasis should lie on inflation or output gap) and other global factors. Changes in the conduct of monetary policy (changes in instruments or inflation targets) are also taken into account.

Quantitative methodologies

Quantitative analysis plays an increasingly important role – but always with a good dose of expert judgement and qualitative appraisal. Key models used are:

• NiGEM DSGE model that makes quarterly GDP projections for many economies 10-years out

• Inflation projections are based on an Augmented Phillips curve model (inputs are fx rates, commodity price forecasts, output gap, long-term inflation expectations)

• A Taylor-rule framework to estimate/gauge the ‘optimal’ central bank policy rate given our economic and inflation projections. The amount of excess liquidity in the system is a key variable in our ‘money market model’ for €STR (formerly Eonia). The relation between excess liquidity and overnight rates follows an S-curve, and we have documented this approach in a special in August 2013: “Excess liquidity and the overnight rate”. However, in recent years, the massive use of quantitative easing and huge liquidity injections in the financial system has made this framework less useful. Benchmark changes (and methodology) for overnight rates and Euribor have also forced us to rely more on qualitative approaches.

• Forward projections for key maturities in money markets are derived from a VAR model that takes into account the historical behaviour of spot rates under certain scenarios for the overnight rate.

Frequency

Updates are usually done on a weekly basis. Update frequency is relatively high but most often it is just a fine-tuning of near-term projections.

Longer-term projections are mostly adjusted on a monthly or quarterly basis. Each quarter there is a thorough assessment of our projections (also in a broader context) where we look at the consistency between countries and each month there is also an additional check of our projections (just before the publication of our Monthly outlook).

Latest update 27-11-2020

Qualitative methodologies

There is no one approach when it comes to how we come to our forecasts or our trade recommendations. This is owing to the fact that the drivers of the rate markets themselves are not constant. For example, in previous years, investors would have assumed that safe haven fixed income markets (i.e Bunds and USTs) might underperform in an environment in which equities are performing well. In the post crisis period, however, both safe haven and risky asset markets have tended to take their cue from expectations as regards policy stimulus with central bank largesse resulting in a hitherto unprecedented positive correlation between these asset classes.

Given the shifting nature of correlations between differing assets, we aim to spot changes in trends/relationships before they are widely appreciated. One clear example here was provided by peripheral debt markets which began to trade in line with equities as of the beginning of 2016 having previously been insulated from changes in global risk appetite in the months following the ECB’s announcement of QE. This, in turn, fits within a bigger picture we developed in the early part of 2016 which was an apparent loss of confidence in central banks’ ability to support risky asset valuations through unconventional policy stimulus. As can be discerned from these examples, our rate views/trading strategies are derived from a framework that can best be described as “macro-thematic”. Of course, by its very nature, this framework is hard to define as the themes which underpin it are in a constant state of flux. (Crucially, the thinking behind our thematic approach is constantly updated and made public via our daily morning notes and our more in-depth weekly publications).

Having spotted what we believe to be an emergent theme (and, crucially, one which we believe has yet to be fully understood by the market), we might make recourse to technicals either to set entry and exit points for a given trade (i.e. we might look at “prior highs” when considering when we might close a position) and/or refer to relative value (as detailed in the section below) on an historical basis. These technical considerations, though, are purely ancillary and do not inform our thinking as regards trade recommendations/yield forecasts. This owing to the fact that in a market beset by regular structural breaks the past is a poor guide to the future.

Quantitative methodologies

General
The main quantitative tool we use is our in-house “Rich/Cheap” model which measures the relative richness/cheapness of government bonds within the euro area by comparing their valuations with similarly dated bonds issued by other countries in the single currency zone, bonds on the same curve and relative to swap rates. This model reports how many standard deviations a given bond is “cheap” or “rich” relative to a 60-day moving average. The 10 bonds which offer the most relative value using this metric are reported in our weekly publication, Rabo Rate Directions, both as regards comparisons with other bonds on the same curve and vs. similarly dated bonds issued by other euro area governments. We also employ such a model for bonds issued by euro area supranationals and agencies which works according to exactly the same principal.

Detailing the SSA market indices
In the context of what is an increasingly challenging market environment, we have introduced a number of interrelated, proprietary market barometers that we hope will assist investors in tracking the evolution of the European SSA market. We use Rabobank’s own request for quote (RFQ) data to build a number of proprietary market indices, namely, the Market Activity Index, the Market Diffusion Index and the Market Momentum Index. These three metrics rely on ‘number of inquiries’ and ‘volume of trade’ SSA RFQ data, which we have broken down into each of the market sub-sectors, namely; supra-nationals, sub-sovereigns (which are predominantly German Länder) and agencies, and also into tenors across these sectors. By doing so, we believe we have captured a rich and unique view of activity across the European SSA market. With this in mind, we feel these measures will prove useful in assessing prevailing market conditions with a view to assisting investors in their efforts to take positions within an increasingly challenging trading environment.

1. The SSA Market Activity index
This proprietary index is simply the absolute number of RFQs that we have seen through the various digital trading systems used by our SSA trading desk in a specific week (Note: these data do not capture any voice requests that were received). Though our analysis revealed little difference in terms of measuring this activity either through the number of inquiries or volume of inquiries, weekly volume statistics show some outliers, whereby, for example, one request for a large volume of one particular issuer/bond does potentially impact the overall index positioning.

As such, we rely on the total number of inquiries in our index calculation, where the index itself uses a base week of 2015W2 (Week 2) = 100. We believe this allows us to create a clearer view of the trend in demand for the combined universe of SSA bonds traded by Rabobank. When taking into consideration the size of Rabobank’s SSA trading platform –the desk is currently ranked number 1 on Bloomberg in terms of number of tickets traded– we believe these data and the index we calculate provide a reliable overview of broader market activity and trends.

To take the index a step further, we break the Market Activity data into more granular form, examining maturity for both inquiries and volume. To define these measures, “volume” is the aggregate volume of inquiries (total ticket size), whilst “inquiries” is simply the number of RFQs, regardless of their actual size. A breakdown of the data indicates how much of the activity witnessed can be allocated to which particular segment of the market, contrasting whether this was based on volume or number of tickets. Although this is a simple aggregate of the number of inquiries and the total volume of inquiries within each SSA maturity sector, we believe this measure provides a vivid picture as to where the level of interest is vs. the volume of inquiries made across the SSA sector over the course of a trading week.

2. The SSA Market Bias index
The market bias index is designed to show whether there is relatively more buy or sell pressure across buckets and within each of the SSA sub-sectors. It is important to note that this metric differs from the Market Activity Index in that it is a diffusion index where 50 is neutral, anything sub-50 signals selling, and any reading above 50 shows a bias towards buying during the week. The computation is straightforward, where any buy volume is seen as a positive number, and any sell as a negative. The difference between the two is then computed and compared on a relative basis to the total volume in the week. This methodology is identical for the number of inquiries, with the exception that inquiry data has been grouped by counterparty (i.e. the party looking to buy/sell). This way, the number of inquiry data highlights buy/sell bias in terms of the “number of participants” looking to buy or sell, while the volume data signals whether there was any large volume.

3. The SSA Market Momentum index
Finally, our Market Momentum Index attempts to highlight whether combined buying/selling pressure has increased/decreased over time and further qualifies the picture presented in the Market Bias Index above. As such, a richer picture of activity can be drawn when the Market Bias and Market Momentum Indices are considered together. For the Market Momentum Index, we present a period of four weeks per SSA sub-sector, so momentum is examined separately for supra-nationals, sub-sovereigns and agencies. This indicator measures the relative size of a weekly change in each maturity bucket over a four week period with each of these weeks compared to the past three months (12 weeks). Here, a reading of ‘0’ represents the single smallest change seen in the past 12 weeks (i.e. the most ‘negative’ change, or the largest shift towards selling in that particular segment), while a score of 100 signals that that particular weekly change is the largest increase (i.e. shift towards buying) seen in the last 3 months. These scores are an average of the change in both inquiries and volume.

By showing a set of 4 bars per maturity bucket for each market segment, where each bar represents one week, we can provide an indication of whether the market is consistently looking to buy more within a specific segment, whether the increase was a one-off within the observation period, or whether it was a reversal from a previous increase in selling pressure.

Frequency

Our trade recommendations are more the product of inspiration than perspiration and, hence, are not created according to any set timetable. Our rate forecasts are reviewed at least once a month prior to the publication of the Rabo research team’s monthly. We update our “Rich/Cheap” model on a weekly basis prior to the publication of “Rabo Rate Directions”.

Latest update 27-11-2020

Qualitative methodologies

The team does not provide explicit forecasts on debt instruments issued by financial institutions or utilities. Neither buy/hold/sell recommendations nor point or range forecasts for specific sector spread indices are given in our publications.
Implicit recommendations are given following extensive qualitative analysis on three levels:

The first level is bottom-up name analysis, where we track company developments of specific issuers. The focus is here on analysing earning reports, where we discuss our view on earnings, capital generation and the outlook. The impact of other news, such as litigation and management changes, is taken into consideration as well when forming a view on the debt instruments of the company.

The second level is instrument analysis, where we track development of specific debt instruments. For Financials this regards the full debt capital stack of European banks and insurers, ranging from AT1 to securitisations. For utilities, the coverage contains senior and hybrid notes. This analysis mainly consists of monitoring market developments, both in the primary and in secondary market. For Financials specifically, analysis is done on the structure of these instruments, and in case of securitisations and covered bonds, also to the asset pool as the secured collateral.

The third level is top-down analysis, where we focus in the impact of macro-economic, monetary policy and regulatory developments on the names, sectors and instruments we follow. The macro analysis is mainly focused on the Benelux, as it is an important element for the credit strength of the loan book of our core coverage and the revenue generation capability of utilities. Moreover, the rates environment is important to consider as well, especially for the solvency of insurance companies and the tariffs utilities are allowed to charge. The regulatory analysis consists of monitoring and scrutinising new laws and/or proposals and regulation of regulatory bodies. For this reason, we also monitor the political arena very closely on this subject.

Quantitative methodologies

The team does not use specific quantitative models, other than a limited use of cash flow models for securitisations and risk free rates for pension funds, and for utilities bottom-up analysis. The cash flow models, with specific parameters on for example prepayment speeds, are all run through the standard Bloomberg ABS interface. For utilities excel-based tools are used.

Market based information for the debt instruments we cover in Financials, is systematically stored in several databases, which are updated on a daily basis. On the basis of this data, we are able to create specific curves, custom spread indices and create new issuance tables and monitor overall market developments. This information is also used to determine fair value for new issues and to provide estimations for new issuance premiums.

Frequency

Implicit recommendations can be changed on a daily basis, depending on news and other developments.

Latest update: October 29, 2024

Qualitative methodologies

FX is driven by a wide range of variables. These include political and economic news, interest rates and central bank guidance. That said, large and significant volume of FX trading is carried out by systematic accounts which use rule-based trading techniques. These fall into many categories such as momentum systems which may work because investors can be slow to respond to new news. It is important to be aware of how these accounts are likely to trade in certain market environments since flow can have significant value in analysing FX markets. Certain currency pairs in certain conditions will enter a particular phases such as risk-on/risk-off, value, smart beta etc. It is important to attempt to recognise these phases and the shocks or events that may throw them off course. Correlations to other assets can be key. Sometimes a currency does not trade on the back of its country’s fundamentals and instead it can exhibit high correlations to another asset such as a certain commodity. The real value of an FX strategist is knowing which strategy to apply to which market and when.

Quantitative methodologies

In 1983 Meese and Rogoff presented findings that a random walk model is superior to all structural models in forecasting FX rates. The results have provided ample fodder for academics and strategists since. In 2002 Cheung et al found that none of the fundamental models (uncovered interest rate parity, sticky price monetary model etc.) consistently outperforms a random walk. However, Brooks et al in 2001 found that fundamentals such as the current account and portfolio flows are the main determinants of exchange rates while Kilian and Taylor indicted that the predictability of exchange rates improves as the forecast period is lengthened.

Academic evidence suggests that economic variables such as interest rates, prices, money and output do not perform well. Models based on the Taylor rule and net foreign asset flows are better. Moosa and Burns and others have shown that models can outperform a random walk if forecasting accuracy is measured in terms of the ability to predict direction. In practice various models can be of use in FX forecast at certain times. However, the value of any FX predictive model is likely to prove inconsistent over time. Purchasing power parity (PPP) is a useful starting point for FX forecasting though it must be accepted that currencies can be undervalued or overvalued for many years so it is of little short-term value. If a currency is either very over or undervalued it can offer some insight into central bank behaviour.

Absolute PPP suggests that the level of the exchange rate will be that which equalizes the levels of prices across various countries – this is often known as the Big Mac measure.

Relative PPP suggests that a currency associated with a higher inflation rate is expected to depreciate vs. a lower inflation rate currency. Inflation reduces the real purchasing power of a currency. Relative PPP relates the change in two countries inflation rates (from a base period of stable price pressures) to the change in their exchange rate. The implication is that the FX rate will compensate for the change in the inflation differential.

Uncovered interest rate parity may not be statistically significant but can offer direction. Uncovered interest rate parity implies that the expected return on a domestic asset will equal the exchange rate-adjusted return on a foreign currency asset. It can be adjusted for forward rates (covered interest rate parity) and for real interest rates. Expectations of how returns can change can therefore be a key determinant of assets prices. Liquidity and ‘riskiness’ of assets must also be taken into consideration.

FX Quant Models

We often refer to quantitative models for forecasting currencies. There are usually three components to these:

1. Fundamental. Above we have explained some issues with these models in the context of predicting movements in FX. However, factors such as inflation indices, current account positions and foreign debt can be influential factors.

2. Technical (see below), these include momentum indicators and key support and resistance levels.

3. Market driven factors include volatility skews, options positioning, and correlations to other assets.

Proprietary models

We have used heat maps to examining the fundamentals of currencies within the same risk group. We determine z-scores for each variable (current account/GDP ratio, CPI inflation rate, debt ratios etc.) and use these to determine which countries are most exposed.

Fundamentals determine which currencies behave as safe havens. In an ideal world a safe haven currency would have both a current account and budget surplus. There would be strong levels of liquidity a coherence central bank and high levels of trust and stability in government and legal systems. A safe haven currency will have significant impact of the behaviour of the central bank in that country.

Technicals

Technicals assume that FX trends can be exploited. Moving averages and measures of moving average convergence divergence (MACD) can be easily followed.

Fibonacci extension: When there is no previous important top as a point of reference for the buyers, Fibonacci extension allows an analyst to obtain higher targets as shown on the monthly USD/ZAR chart. The 76.4% Fibonacci retracement level tends to provide a relatively solid support. Risk/reward is skewed in favour of establishing long positions.

Patterns: Price action can form various patterns which can have predictive qualities. For example, a break higher from a flag can indicate a new phase of an upside trend.

Head and shoulders is a famous reversal pattern. By measuring the distance between the head and the neckline, we obtained a level as a potential target on the downside. That said, we have to take into consideration that Fibonacci retracement levels will provide support.

Charts can reveal that some currency pairs are capable of producing a certain rhythm. For example: after a multi-month rally, USD/TRY tends to pause and consolidate its gains before another leg higher resumes.

According to the Elliott wave principle, the markets tend to move in waves. A long-term trend is formed by five waves with the wave three the longest.

Frequency

FX forecasts are checked at least once a week and more frequently when a market events dictate.

Latest update: November 16, 2024

Credit strategy & regulation

Regular publications concerned: Asset-backed morning comment, Financials Daily, Financials Weekly, Focus on ABS, Focus on Covered Bonds, Focus on SSAs, Housing Market Bulletin, Bank Bulletin, Insurance Bulletin, Regulation Update and Focus on Dutch Pension reform and regulation.

Qualitative methodologies

The team does not provide explicit forecasts on debt instruments issued by financial institutions. Neither buy/hold/sell recommendations nor point forecasts for specific sector spread indices are given in our publications. We do however provide range of spread forecasts.

Implicit recommendations are given following extensive qualitative analysis on three levels. The first level is bottom-up name analysis, where we track developments of specific issuers. The focus here is on analysing earning reports, where we discuss our view on earnings, capital generation and the outlook. The impact of other news, such as litigation and management changes, is taken into consideration as well when forming a view on the debt instruments of the company.

The second level is instrument analysis, where we track development of specific debt instruments. For Financials this regards the full debt capital stack of European banks and insurers, ranging from AT1 to securitisations and covered bonds. This analysis mainly consists of monitoring market developments, both in the primary and secondary market. For Financials specifically, analysis is done on the structure of these instruments, and in case of securitisations and covered bonds, also to the asset pool as the secured collateral.

The third level is top-down analysis, where we focus in the impact of macro-economic, monetary policy and regulatory developments on the names, sectors and instruments we follow. The macro analysis is mainly focused on the Benelux, as it is an important element for the credit strength of the loan book of our core coverage. Moreover, the rates environment is important to consider as well, especially for the solvency of insurance companies. The regulatory analysis consists of monitoring and scrutinising new laws and/or proposals and regulation of regulatory bodies. For this reason, we also monitor the political arena very closely on this subject.

Quantitative methodologies

The team does not use specific quantitative models, other than a limited use of cash flow models for securitisations and risk free rates for pension funds. The cash flow models, with specific parameters on for example prepayment speeds, are all run through the standard Bloomberg ABS interface. For SSAs, we rely on market pricing from Bloomberg.

Market-based information for the debt instruments we cover in Financials, is systematically stored in several databases, which are updated on a daily basis. On the basis of this data, we are able to create specific curves, custom spread indices and create new issuance tables and monitor overall market developments. This information is also used to determine fair value for new issues and to provide estimations for new issuance premiums.

Based on public data of the regulators we have created a dashboard which allows us to identify key trends in the Dutch pension sector.

Frequency

Implicit recommendations can be changed on a daily basis, depending on news and other developments.

Last update: October 28, 2024

Rabobank agri commodity markets price forecasts are released in the ACMR monthly or in special reports if needed to the full agri commodity markets distribution list. Price forecasts are limited to forecasts of major exchange traded agri commodities including wheat (CME Chicago, Kansas, Minneapolis and Euronext), corn (CME Chicago, Euronext), soybeans (CME Chicago), soybean oil and meal (CME Chicago), palm oil (MDE-Bursa) sugar (ICE #11 and ICE White sugar), cocoa (ICE NY, ICE London), coffee (ICE Arabica, ICE Robusta) cotton (ICE #2).

The forecasts reflect the expected average exchange price for the quarters about 1 year in the future and an indication for a quarter two years in the future.

Price forecasts are based on fundamental agri commodity analysis combined with technical analysis, managed money positioning, spot FX and Rabobank exchange rate forecasts.

Fundamental analysis is performed for major export and import regions, aggregated up to a global level on a crop marketing year basis and use when available official reports as a historic baseline. Fundamental forecasts are broadly derived as follows:

• Production forecasts are based on acreage and yield forecasts.

• Acreage forecasts are based on planted acreage expectations, considering farmer margins, short and medium-term weather forecasts, planting progress as well as typical and weather related conversions of planted acreage into harvested acreage.

• Yields forecasts usually start as trend yield projections and are adjusted based on weather (both historic and forecast) and pests and diseases.

• Demand forecasts are based on GDP and population growth, dietary changes, waste and early harvest assumptions, adjusted by official trade and stock reports, combined with official weekly/monthly demand reports.

• Trade flow forecast are based on local supply and demand estimates and official trade flow reports from various sources.

Price forecast models take into account what is already priced in current exchange prices as well as events that can be expected with a high likelihood.

Fund positioning takes into account the figures reported by CFTC, which are also shared with clients in Rabobank’s Commitment of Traders reports.

Technical analysis is based on in-house as well as external analysis.

Our forecasting models include exchange rate forecasts made by Rabobank’s FX strategists, as large volumes of agri commodities are traded globally and exchange rates significantly impact exports/imports of certain regions and thus have a strong impact on the availability and ending stocks of these commodities in different countries and therefore also reflect back into futures prices at the above mentioned exchanges.

The price forecasts also include political or other uncertainties and risks and are adjusted to reflect these factors. The most important risk/uncertainty factors are usually described in the monthly reports together with the price forecasts and are largely referred to in the report in which the price forecast is published.