Macro-quantamental scorecards are condensed visualizations of point-in-time economic information for a specific financial market. Their defining characteristic is the combination of efficient presentation and evidence of empirical power. This post and the accompanying Python code show how to build scorecards for duration exposure based on six thematic scores: excess inflation, excess economic growth, overconfidence, labour market tightening, financial conditions, and government finance. All thematic scores have displayed predictive power for interest rate swap returns in the U.S. and the euro area over the past 25 years. Since economic change is often gradual and requires attention to a broad range of indicators, monitoring can be tedious and costly. The influence of such change can, therefore, build surreptitiously. Macro-quantamental scorecards cut information costs and attention time and, hence, improve the information efficiency of the investment process.
The post below is based on Macrosynergy’s proprietary research.
Please quote as “Andresen, Sam, Regis, Glenn and Sueppel, Ralph, ‘Macro-quantamental scorecards: A Python kit for fixed income markets,’ Macrosynergy research post, August 2024.”
The attached Jupyter notebook allows for the audit and replication of the research results. The notebook’s operation requires access to J.P. Morgan DataQuery to download data from JPMaQS, a premium service of quantamental indicators. J.P. Morgan offers free trials for institutional clients.
Also, an academic research support program sponsors data sets for relevant projects.
The concept of macro-quantamental scorecards
Macro-quantamental scorecards are condensed visualizations of a financial market’s relevant economic information space. They are based on quantamental indicators, i.e., point-in-time information states of standard macroeconomic data, such as inflation or economic growth rates. Quantamental indicators are based on historic vintages of economic data and can be downloaded from the J.P. Morgan Macrosynergy Quantamental System (JPMaQS). Quantamental indicators can be condensed by normalization and logical combination into thematic scores. A macro-quantamental score here means the value of an economic indicator relative to its neutral level in standard deviations for a group of similar countries up to the concurrent point in time. A thematic score is a weighted average of indicator scores that inform on the same economic concept, typically re-normalized so that the composite thematic score is expressed in standard deviations.
Quantamental indicators in a scorecard should be based on theory and empirical evidence. For example, low unemployment is expected to hurt long-duration positions in the fixed-income market because it reduces monetary policy accommodation and bodes for higher wage and price pressures. Moreover, there should be a predictive relation in the presence of rational inattention. Predictive power can easily be tested by checking the direction and significance of the link between point-in-time macroeconomic information states and subsequent long-duration returns. This ability to verify historic success easily naturally fosters a focus on relevant macro information. It is the special characteristic that sets macro-quantamental scorecards from other forms of macroeconomic heatmaps. In summary, macro-quantamental scorecards serve two purposes:
- Easy and efficient visualization of key economic developments relevant to market trends and risk, typically in the form of thematic scores.
- Accompanying empirical evidence of the predictive power of individual indicators or thematic scores in the past.
The benefit of this type of information management is consistent and low-cost monitoring of a broad range of relevant macro factors. It is superior to standard information ingestion in discretionary, which is characterized by spells of inattention, bouts of mental excitement, and broken links with empirical evidence. Moreover, visualized scores can easily be implemented as a systematic default strategy. Default here means strategic positioning bias, which works comfortably with overlays of other systematic strategies or trade ideas.
Macroeconomic information space
This post presents simple macro-quantamental scorecards for the U.S. and euro area rate markets. The target contract is a five-year interest rate swap fixed receiver as a proxy for duration risk. Scorecards are built by the principle of “simplest plausible version”. They use popular indicators in their conventional forms. Such basic scorecards can be enhanced by using many more indicators of JPMaQS and should be adapted to the priorities of a specific trading book.
Regarding interest rate swaps, we characterize the macro environment through six thematic macro-quantamental scores, all composites of underlying quantamental categories. For details on the underlying categories, see Annex 1 below. Broadly, the thematic scores can be labelled as follows:
- Excess inflation: This thematic score is based on various metrics of short—to medium-term consumer price trends relative to the central bank’s inflation target. Excess inflation bodes for tighter monetary policy and higher inflation risk premia and, hence, puts upward pressure on long-dated rates.
- Excess economic and demand growth: This thematic score combines higher-frequency GDP growth estimates and measures of domestic demand over and above medium-term economic growth. Excess growth and demand support tighter monetary policy and higher real interest rates.
- Excess confidence: This theme is based on a composite of business and consumer survey indicators normalized based on conceptual and theoretical neutral levels and historical volatility. High consumer and business sentiment bodes for stronger aggregate demand and monetary policy tightening, and, hence, higher real rates.
- Labour market tightening: The thematic score combines declining unemployment, excess employment growth and excess wage growth. Tightening labour markets put upward pressure on the inflation outlook, real monetary policy rates, and long-term yields.
- Financial conditions: This score is based on real interest rates and private credit growth. High real rates and credit demand indicate elevated risk or term premia and augur well for long-term rate receiver returns.
- Government finance: This theme is a composite of an excess government balance and borrowing requirement. Strong government balances and low borrowing lead to low supply pressure and, hence, support bond and duration returns.
The thematic scores are computed in three steps.
- All underlying indicators are calculated as differences from a plausible normal or neutral level and given the sign that, according to standard theory, aligns them positively with duration returns. For example, in the case of inflation, we subtract an effective inflation target and put the difference in negative terms. See Annex 1 below for details.
- We perform point-in-time panel-based normalization, i.e., divide information state values for any date by their historic average for all countries and cap absolute values at 3 standard deviations to de-emphasize outliers.
- We average all indicator scores that belong to a theme and re-score their values, i.e., perform a second round of point-in-time normalization to compensate for the diversification effects of combining different indicators.
The important point is that all values have similar meanings and scales and are easily interpreted at a glance.
A snapshot of condensed macro factors
The ScoreVisualizers class of the Macrosynergy package supports easy display of standard scorecards. The class scores or re-scores indicators calculate composites, produce single point-in-time snapshots, and visualize longer-term histories of period averages across countries or indicators.
The small scorecard below shows the state of all macro factor scores for the U.S. and the euro area, as well as a simple (re-scored) average. Simply put, blue is good for duration returns, red is bad, and all numbers are standard deviations. The signs of the composite scores, or their underlying indicators, are always set so that a higher value is better for duration returns, according to mainstream theory. For the present thematic scores, this mostly requires putting them in negative terms.
The composite is simply a re-scored average of the thematic scores. For simplicity, we have not assigned weights across themes or optimized themes. At the time of completion of this post, the scores added up to a positive macro environment for duration exposure for both the U.S. and the euro area. The macro support was not quite uniform across groups. While low confidence, moderating inflation, and tighter financial conditions augured well for duration exposure in the U.S., tighter labour markets and weak government finances look more like headwinds.
A history of macro factor scores (U.S.)
Normally, thematic macro-quantamental scores change gradually rather than abruptly. Without shocks and crises, they can even pass through multi-year seasons. In conjunction with the multitude of information types, this gradual evolution appears “boring” and discourages continuous systematic attention. This corresponds to anecdotal evidence that economic trends often proceed surreptitiously and do not make popular headline news unless they are associated with crises or until their effects have become obviously consequential for people’s lives.
The below scorecard illustrates the history of thematic macro scores in the U.S. over the past 25 years. Scores are annual averages, except for the latest column and the 2024 average, which is January to July. Based on the composite score, macro support for duration exposure was stronger in the 2000s than in the 2010s. In the 2020s, macro support for duration collapsed in the wake of the pandemic but recovered recently. Generally, macro conditions have displayed drastic swings in the runup and wake of crises and recessions while evolving gradually in other times.
Recent quarters saw a marked shift from strong macro headwinds to slight tailwinds for duration risk. This shift owned mostly to three forces:
- inflation pressure surged after COVID and then subsided relative to higher perceived effective targets,
- economic confidence had rebounded in the post-COVID recovery and then fell back, and
- financial conditions shifted to the restrictive side in the wake of inflation concerns and related monetary tightening.
Annex 2 below details the evolution of all sub-scores of all themes. The constituents of a thematic score are not always closely aligned. For example, labour markets may be tightening according to employment and unemployment trends but still easing according to wage trends, with the latter being subject to special structural developments and inflation expectations.
Empirical evidence of the relevance of scores
The hypothesized directional relations of the above macro factors and duration returns (as proxied by returns on 5-year interest rate swap fixed receivers) follow standard economic theory, such as the Taylor rule of monetary policy or New Keynesian economics. A condensed version of the latter can be found here. The presumed predictive power of all these scores is based on the principle of rational inattention of market participants and the prevalence of implicit subsidies in financial markets. This post’s scores and composite are not optimized for predictive power.
The graphs below check whether the six chosen thematic macro scores predicted subsequent 5-year IRS receiver returns in the U.S. and the euro area at a monthly frequency. The basis for the significance of this panel relation is the Macrosynergy panel test, which is explained here.
All predictive relations have been positive, i.e. in the theoretically proposed direction. Moreover, the forward correlation of excess growth, excess confidence, labour market tightening, and financial conditions have all been highly significant. Excess inflation and government finances have been less significant. However, this does not disqualify their consideration. Even a 25-year sample is strongly affected by temporary macroeconomic regimes or seasons. Indeed, neither inflation nor public finances have been a great concern for the U.S. over the past three decades. Yet the experiences of the 1970s and 1980s and the last two years have demonstrated that the importance of inflation for fixed-income markets increases with price growth and challenges to the central bank’s target. Also, in the 8-country example below, the government finance score has been a significant predictor over the last 25 years.
The predictive power of a simple average macro score with respect to subsequent 5-year IRS returns has been strong and highly significant at both monthly and quarterly frequencies. This gives investors confidence that macro matters plausibly and predictably. While the individual factors’ forward correlations may be partly seasonal and reflect specifics of the past decade, the forward correlation of a broad non-optimized composite is a plausibly persistent feature.
A broader global scorecard
The snapshot and historical scorecard below take a broader, more global view, looking at the macro scoring of eight major developed fixed-income markets. This broadens visualization geographically and changes the basis of normalization. This means the panel data set used for calculating standard deviations in this context comprises eight countries.
The latest eight-country snapshot confirms the positive message of the U.S. and euro area data. Indeed, the average developed market environment for rates looks more positive than that for the U.S. alone, reflecting a rare constellation of positive scores for all countries. The exception is Japan, which scores highly negative in the field of financial conditions.
Most factors post similar values across countries, but this is not uniform. Labour market signs are different across currency areas. Also, government finances are generally a differentiating factor.
Macro conditions across countries have revealed strong global commonalities over the past 25 years. Diversity has been more common in “quieter” times than in pronounced financial economic cycles.
A global weighted history
Beyond cross-country history, a final useful perspective is global scores. Global thematic scores are weighted averages of country scores, with point-in-time series of nominal GDP serving as weights. A case can be made that such global rather than local conditions dominate fixed-income markets.
Global information states of excess growth, inflation, confidence, and labour market tightening all display pronounced cyclical patterns around a stationary trend. The cycles are not always synchronized, however. For example, at the outset of the great financial crisis 2008, excess inflation pressure detracted from the performance of duration positions, while business sentiment and estimated GDP growth already supported them. And labour market scores lagged output and confidence scores. Such observations explain why it is justified to consider a range of relevant macro scores, even if they are positively correlated.
Financial conditions and government finance scores have shown secular trends and cycles. Simply put, low real interest rates and rising government deficits have jointly eroded expected duration returns over time, or at least for a large part of the last 25 years. This reflects realistically that the macro environment can be subject to medium-term shifts, often overlooked by traders with short-term horizons.
Annex 1:
Categories of information states used for the various macro scores
Excess inflation (presumed negative impact):
Excess core CPI inflation. This is based on the core consumer price index preferred by the central bank. The basic idea behind core price indices is to exclude volatile or erratic prices that cloud the detection of a trend (view documentation). Core inflation is measured as % over a year ago and % change of the last 3 months over the previous three months, seasonally adjusted annual rate. We subtract from this the effective inflation target of the central bank (view documentation).
Excess headline CPI inflation. This is the standard annual headline consumer price inflation, measured as a % change over a year ago (view documentation). We subtract from this the effective inflation target of the central bank (view documentation).
Excess 1-year ahead inflation expectations. The estimate follows a formula that assumes that market participants form their inflation expectations based on the recent inflation rate (adjusted for jumps and outliers) and the effective inflation target. For recent inflation, we use, where available, an average of headline and core inflation. For the 1-year forward horizon, the weight of recent inflation to the effective target is 3/4 to ¼ (view documentation). We subtract from this the effective inflation target of the central bank (view documentation).
Excess growth (presumed negative impact):
Excess “intuitive” GDP growth. This is the latest estimable GDP growth trend, % over a year ago, 3-month moving average, based on actual national accounts and monthly activity data, based on sets of regressions that replicate conventional charting methods in markets (Macrosynergy methodology). It is based on all major activity indicators and GLS regression (view documentation). From this, we subtract a 5-year real GDP trend (view documentation).
Excess “technical” or “nowcast” GDP growth. This is estimated based on supervised learning of nowcasters for real GDP growth trend, % over a year ago, and 3-month moving average. Growth rates are estimated for every new release of a set of credible macroeconomic predictors using their actual or estimated available data vintages at the time (view documentation). From this we subtract a 5-year real GDP trend (view documentation).
Excess real private consumption growth. This is based on either monthly data or quarterly national accounts and measures real household consumption as % over a year ago in 3-month moving averages or quarterly (view documentation). From this, we subtract a 5-year real GDP trend (view documentation).
Excess nominal import growth. This is an indicator of merchandise imports in local currency, adjusted for seasonal effects, working days and holidays (view documentation). The growth rate is the increase of the last 6 months versus the previous 6 months as an annualized rate, thus containing the volatility of the series. From this, we subtract a 5-year real GDP trend (view documentation) and the effective inflation target (view documentation).
Excess confidence (presumed negative impact):
Manufacturing confidence score. This is a sequentially normalized and general score of relevant industry surveys, seasonally adjusted and as a 3-month moving average (view documentation).
Consumer confidence score. This is a sequentially normalized and general score of relevant household surveys, seasonally adjusted and as a 3-month moving average (view documentation).
Services confidence score. This is a sequentially normalized and general score of relevant broad services surveys, seasonally adjusted and as a 3-month moving average (view documentation).
Construction confidence score. This is a sequentially normalized and general score of relevant construction business surveys, seasonally adjusted and as a 3-month moving average (view documentation).
Labour market tightening (presumed negative impact):
Excess employment growth: This is payroll growth in % over a year ago, 3-month moving averages (view documentation) minus annual workforce growth over the past 5 years (view documentation).
Excess unemployment rate: This is the main unemployment rate minus a 5-year moving average, in %-points of the workforce, also called the unemployment gap, in negative terms. (view documentation).
Excess wage growth. This is defined as the main local wage growth measure per unit of output in excess of the effective estimated inflation target. Technically, excess wage growth refers to the increase in wages relative to the growth in productivity or output beyond what is considered consistent with the targeted level of inflation (view documentation).
Financial tightness (presumed positive impact):
Real short-term interest rate. This is the main 1-month money market rate or the closest proxy in the local currency area (view documentation).
Real 5-year swap yield. This is calculated as the main local interest rate swap yield minus formulaic 5-year ahead inflation expectations (view documentation).
Excess private credit growth. This is the % change in private bank credit over a year ago and as a 3-month moving average (view documentation) in negative terms. From this we subtract a 5-year real GDP trend (view documentation) and the effective inflation target.
Government financial strength (presumed positive impact):
Excess general government balance. This refers to the difference between expected general government revenues and expenditures for the current fiscal year, % of GDP (view documentation). Expenditures include interest spending, i.e. they are affected by debt servicing costs. It becomes an excess indicator by removing 3%, which is the normal deficit ratio according to the old Maastricht threshold.
Excess net government borrowing (negative). Government borrowing requirements are generally defined as the financial needs of a government when calculated via a cash accounting basis. It is measured as a 1-year moving average (as monthly and quarterly values are very volatile) and as % of GDP (view documentation). It becomes an excess indicator by adding 3%, the normal deficit ratio according to the old Maastricht threshold.
Annex 2:
Evolution of U.S. thematic scores and their constituents since 2000
