Macro factors and sectoral equity allocation

Jupyter Notebook of factor calculation Jupyter Notebook of factor analysis

Returns of major equity sector indices relative to the overall market plausibly depend on macroeconomic trends. Certain economic developments, such as the state of the business cycle, relative price trends, or financial conditions, drive divergences in business conditions. We test the predictive power of plausible point-in-time macro factors for the relative performance of the 11 major equity sectors in 12 developed countries over an almost 25-year period since 2000.
While not all plausible simple macro hypotheses are supported by the evidence, “conceptual parity scores” that simply average all (normalized) factors have displayed significant predictive power for relative returns of most sectors. The joint risk-adjusted returns generated by relative allocation across all 11 sectors are sizable, with a Sharpe ratio of over 1. This suggests that macro factor-based allocation may more than double the risk-adjusted returns of standard equity portfolios.

The post below is based on Macrosynergy’s proprietary research.
Please quote as “Rushil and Sueppel, Ralph, ‘Macro factors and sectoral equity allocation,’ Macrosynergy research post, June 2024.”

Jupyter notebooks allow audit and replication of the research results. A notebook for factor calculation can be downloaded here. A notebook for subsequent factor analysis can be downloaded here. The notebooks’ 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.

This post ties in with this site’s summary of macro trends and systematic value.

Why macro matters for cross-sector equity performance

Equity sectors are characterized by commonalities in business models. One of these commonalities is their dependence on macroeconomic conditions, such as the business cycle, inflation, debt servicing costs, or real currency strength. Put simply, earnings prospects and risk premia of different sectors respond differently to macroeconomic developments. Moreover, judging from applied research and public discussion, the equity market is far from efficient in accounting for these macroeconomic influences.

For the analysis below, we will examine the following sectors in accordance with the “Global Industry Classification Standards” (GICS) developed for the equity markets in 1999 jointly by MSCI and Standard & Poor’s.

1. Energy: The sector comprises companies that support the production and transformation of energy. There are two types. The first type focuses on exploring, producing, refining, marketing, and storing oil, gas, and consumable fuels. The second type provides equipment and services for the oil and gas industries, including drilling, well services, and related equipment manufacturing.
2. Materials: The sector encompasses a wide range of companies engaged in discovering, developing, and processing raw materials. These include chemicals, construction materials (such as cement and bricks), container and packaging materials (such as plastic and glass), base and precious metals, industrial minerals, paper, and other forest products.
3. Industrials: The sector contains a broad range of companies involved in producing goods used in construction and manufacturing (capital goods) as well as providing commercial services and transportation. The area of capital goods includes aerospace and defense, building products, construction and engineering, electrical equipment, industrial conglomerates, and machinery. The commercial services sub-sectors include waste management, office supplies, security services, and professional services (consulting, staffing, and research). The transportation area includes air freight and logistics, airlines, marine transportation, road and rail transportation, and transportation infrastructure companies.
4. Consumer discretionary: This sector comprises companies producing consumer goods and services considered non-essential but desirable when disposable income is sufficient. The main areas are automobiles, consumer durables, apparel, consumer services (such as hotels and restaurants), and various retail businesses.
5. Consumer staples: This sector includes companies that produce and distribute presumed essential consumer products that households purchase regardless of economic conditions. These products mainly include food, beverages, household goods, and personal care items.
6. Health care: The sector includes companies that provide medical services, manufacture medical equipment, or produce drugs. It has two main areas. The first features health care equipment and services. It includes manufacturers of medical products and supplies, providers of health care services (such as hospitals and nursing homes), and companies that provide technology services (such as electronic health records). The second area features research, development, and production of pharmaceuticals, biotechnology, and life sciences tools.
7. Financials: This sector provides financial services, including banking, investment services, insurance, and financial technology (fintech). The four main subsectors are banks, diversified financials (such as asset management, credit cards, and financial exchanges), insurance, and investment trusts.
8. Information technology: This sector includes companies that produce software, hardware, and semiconductor equipment, as well as those that provide IT services.
9. Communication services: This sector features companies that broadly provide communication services and entertainment content. It contains two main areas. The first is telecommunication services, which provide the means for telecommunication, including traditional fixed-line telephone services, broadband internet services, and wireless telecommunication services. The second area is media and entertainment, which focuses on the creation and distribution of content for broadcasting, home entertainment, movies, music, video games, social media platforms, search engines, and so forth.
10. Utilities: This sector includes companies that provide essential utility services such as electricity and water. Their activities include generation, transmission, and distribution, and they are typically subject to tight regulations. Standard classification distinguishes five types of utilities: electric utilities, gas utilities, water utilities, multi-utilities, and independent power and renewable electricity producers.
11. Real estate: This sector focuses on real estate development and operation. It encompasses property ownership, development, management, and leasing. It also includes Real Estate Investment Trusts (REITs) that invest in various property types.

Equity sector return data

The target returns of this analysis are sectoral cash equity excess returns, approximated as the difference between the returns of one sector versus an unweighted average of all sectors. Sectoral return data are available on JPMaQS (view documentation here) for 12 economies, which are (alphabetically by currency symbol): Australia (AUD), Canada (CAD), Switzerland (CHF), the euro area (EUR), the UK (GBP), Israel (ILS), Japan (JPY), Norway (NOK), New Zealand (NZD), Sweden (SEK), Singapore (SGD), and the U.S. (USD). The underlying equity return data comes from the J.P. Morgan SIFT database. SIFT stands for Strategic Indices Fundamental Toolkit and is a dataset comprising pricing and fundamental data for 20,000+ equity indices. The data are categorized by region/country, market size segment, sector/industry group, and different factors, including the overall market.

The below shows relative sectoral returns for the U.S. and the euro area as examples. The broad patterns of sectoral performances are similar and, in both cases, illustrate that relative sectoral returns have both long-term trends and cyclical patterns.

Macro-quantamental factor candidates

The predictors of relative equity sector returns are macro indicators in a point-in-time format. These are called “macro-quantamental indicators” and can be taken from the J.P. Morgan Macrosynergy Quantamental System (“JPMaQS”). Unlike regular economic time series, their values are based solely on information available at the time of recording. Consequently, macro-quantamental indicators can be compared to market price data and are well-suited for backtesting trading ideas and implementing algorithmic strategies.

For the prediction of sectoral equity returns, we use a set of macro-quantamental categories (indicators across all 11 equity countries) to express or calculate very simple, plausible sectoral equity factors with neutral zero values:

• Excess output growth: For each economy, we take differences between various production growth trends, as % over a year ago and 3-month moving average, and the past five years’ median GDP growth (view documentation). The output growth types are “technical” GDP growth, i.e., a form of point-in-time nowcast (view documentation), industrial production growth (view documentation), and construction growth (view documentation). Information states of excess growth trends are indicative of past economic performance but also of risks for monetary policy tightening, high inventories, and cyclical downturns.
• Excess private consumption and retail sales growth: These are differences between household spending growth, as % over a year ago and 3-month moving average, and medium-term GDP growth. The first metric is real private consumption (view documentation) minus medium-term GDP growth. The second metric is nominal retail sales growth (view documentation) minus the sum of medium-term GDP growth and the effective inflation target. For Israel and Singapore, nominal retail sales are approximated by the sum of real sales and headline inflation. The third metric is real retail sales growth (view documentation) minus medium-term GDP growth. Excess consumer spending trends are indicative of past sales but also of monetary policy tightening risks, excess inventories, and dangers of cyclical downturns.
• Excess export growth: These are local-currency export growth (view documentation), as % over a year ago and a 3-month moving average minus estimated medium-term nominal GDP growth. The latter is approximated as the sum of the 5-year rolling median of real GDP growth and the effective inflation target. Excess export growth in local currency terms is indicative of the profitability of tradable goods industries.
• Labor market indicators: Excess employment growth is the difference between employment growth (view documentation), as % over a year ago, and workforce growth (view documentation). An alternative perspective is the change in the unemployment rate over a year ago (view documentation) and the change in the unemployment rate relative to a five-year moving average (view documentation). High employment growth and declining unemployment support consumer spending, particularly of low-income households, and are beneficial for households’ financial position.
• Business survey scores: For each economy and to the extent that surveys are available, equity sector factor research uses seasonally adjusted scores for manufacturing companies’ confidence (view documentation), construction businesses’ confidence (view documentation), services business confidence (view documentation), as well as short-term seasonally-adjusted changes of all the aforementioned indicators, typically the difference of the past three month’s confidence level versus the previous three months. The level of confidence is often related to the state of the business cycle and indicates the recent performance of the respective sector. Short-term changes in confidence should be indicative of changes in business conditions.
• Excess private credit growth: For each country we take the difference between private credit growth (view documentation) , as % over a year ago, and estimated medium-term nominal GDP growth. Excess credit growth is indicative of business expansion in the financial sector and can be both cyclical and structural.
• Excess broad inflation: For each currency area, we subtract an effective inflation target from standard metrics of headline CPI, core CPI (view documentation), and PPI inflation (view documentation), as % over a year ago. The effective inflation target is the estimated official inflation target plus an adjustment for past “target misses.”  This adjustment is the average gap between actual inflation and the estimated official target mean for the last three years (vie documentation). Excess local inflation is indicative of the pricing power of consumer goods producers but also puts upside pressure on real interest rates and increases risks of long-duration borrowing and lending.
• Excess food and energy CPI inflation: For each economy we measure food CPI inflation as price growth for unprocessed food, beverages, and tobacco and energy CPI inflation as price growth of transport fuels and household energy (view documentation) Excess inflation is again the difference between these two inflation metrics, as % over a year ago, and the effective inflation target. Excess food or energy inflation is indicative of supply shocks and pricing power in certain sectors, typically at the expense of others.
• Excess real wage growth: For each economy, we calculate the difference between wage growth, as % over a year ago (view documentation), and the sum of a medium-term productivity growth trend and the effective inflation target. Medium-term productivity growth is approximated as the difference between real GDP growth median over the past five years and annual workforce growth (view documentation). High real wage growth should benefit low-income household spending but impair the competitiveness of labor-intensive local producers.
• Net debt servicing ratios: These are real-time measures of household and corporate net debt servicing ratios, i.e., interest spending minus interest income, as a percentage of concurrent nominal GDP (view documentation). High or rising debt servicing ratios bode for deteriorating credit quality in the local financial sector.
• Excess extrapolated government debt ratios: These are extrapolated general government debt-to-GDP ratios in 10 years, given the current primary balance, real interest rate estimates, and concurrent general government debt ratio (view documentation). This ratio minus 100% is a metric of fiscal and debt escalation risk, where fear of default and rising bond yields move into mutually reinforcing dynamics. Sovereign default risk is particularly perilous for the local financial sector.
• Commodity inventory scores: These are information states of seasonally adjusted normalized excess inventory measures for commodities in the U.S., China, and LME warehouses (view documentation). For this post, they are used and aggregated for base metals and major fuels contracts. Conceptually, they are global indicators. High inventory scores are expected to reduce future demand for the stored materials.
• Openness–adjusted real appreciation: Real appreciation, measured here as % over a year ago, means that local currency prices are increasing relative to a trade-weighted average of foreign currency prices (view documentation). Openness adjustment means that the (estimated) rate of real appreciation is multiplied by the ratio of merchandise exports and imports to GDP (5-year moving averages). Real appreciation benefits local purchasing power but tends to undermine profits and competitiveness of local companies that compete internationally.
• Commodity-based terms of trade: Commodity terms-of-trade dynamics are ratios of approximate export and import price dynamics, whereby the latter are calculated on a daily basis based on concurrent vintages of commodity prices (in USD) and commodity trade shares (view documentation). Improving terms of trade bodes well for local purchasing power and the revenues of locally-sourced raw materials businesses.
• Interest rate metrics: For this post, we focus on real 2-year and 5-year interest rate swap yields (view documentation) and the spread between them. High real yields dampen demand for capital goods, durable consumer goods, and housing. A steep (real) yield curve bodes well for the profitability of term transformation in the financial sector.
• Relative refined energy versus crude returns: Based on generic returns in JPMaQS (view documentation), this is a plausible special indicator suggested for the performance of the materials sector.

Categories are calculated for each country, either based on the local data alone (“local”) or as a weighted average of local and global values, whereby the international is assumed to be in accordance with the share of external trade flows in GDP (“weighted global”). The choice between “local” and “weighted global” depends on business focus. For example, materials goods producers mostly have a global perspective, whereas most banks and utilities are more focused on the local market. Also, there are some naturally global variables, such as commodity inventories (“global”).

If a category cannot be calculated for a particular cross-section and period, the below analysis is based on the available data alone. This means that we analyze based on different and unbalanced panels across categories. This correctly replicates the actual availability of data in real-time in the past. All factor candidate categories are normalized sequentially, i.e., divided by the average absolute distance of their values from their zero “neutral” levels across the panel, up to the date to which the normalization is applied. The scores are also winsorized (“clipped”) at two standard deviations to reduce the influence of outliers.

Macro factor hypotheses and empirical tests in general

We have chosen a set of six hypotheses of macro factors of sector return outperformance for each of the 11 equity sectors. The factors are formulated so that their theoretical impact on the relative sector return versus market return should be positive, and a zero value should be neutral for the target return. They are listed in the top table of each section. The hypotheses are rough, and no attempt has been made to refine the factors to maximize predictive power. Similarly, the selection of the six factors for each sector has been based on simplicity and plausibility rather than a formal optimization process. This post aims to see how much value can be generated with common sense and betting on market inefficiency rather than assessing a special formal selection method.

As a result, many hypotheses have not been backed empirically for the past 25 years. This means they either have a wrong sign or low significance. However, macro factors are notoriously seasonal, and even 2-3 decades may be subject to a specific environment and may not reveal all important structural relations. Thus, to assess the general value of common-sense macro factors, we calculated an average of all six hypothesized macro factors for each sector. Since the factors are normalized, a “conceptual parity score” is obtained. Conceptual parity means giving each factor equal weight, irrespective of cross-factor correlation and past predictive power. Then, we assess the significance of the predictive power of a conceptual parity score at the end of one month for the relative sectoral return of the next month and the economic value of the score based on a “naïve PnL”:

• Significance of predictive relations: We have tested month-ahead predictive power for all 12 markets (currency areas) over almost 25 years, called a panel. Statistical tests are based on a special panel regression model that adjusts the features and targets for common global influences across countries (view post here). We call this the “Macrosynergy panel test,” but under the hood, it uses the coefficient significance statistic of a period-specific random effects model. For test validity, it is important that the hypothesized relation between features and targets is similar across countries and that the country-specific features matter. For the purpose of this post, we will call a panel relation with more than 90% probability significant.
• Economic value of sectoral signals according to naïve PnLs: We calculate profit and loss series through a standard procedure that takes positions in accordance with normalized signals and regular rebalancing at the beginning of each month, in accordance with signals at the end of the previous month, allowing for a 1-day time-lapse for trading. The trading signals are capped at a maximum of two standard deviations as a reasonable risk limit. A naïve PnL does not consider transaction costs or risk management. It is thus not a realistic backtest of actual financial returns in a specific institutional setting. However, it is an objective and undistorted representation of the economic value of the signals.

The joint naïve PnLs of the 11 relative sectoral equity strategies indicate the material economic value of using macro factors for allocation strategies. However, they must be taken with two important caveats at this stage.

1. The selection of factor candidates is influenced by hindsight, even without optimization. Factor selection is based on judgment and “conventional wisdom,” which naturally reflects experiences and observations of past decades. There is no guarantee that we would have chosen the same set of factors 25 years ago. On the other side, there are many valid hypotheses that we have not considered in this post for the sake of simplicity and speed. We will deploy sequential learning in a future post to derive more objective factor selection and signal generation.
2. The implementation of relative sectoral trades may be limited by the size of markets and short-sale restrictions. The below-simulated value generation assumes that all long and short positions can be taken equally across countries and that there is no limit on cumulative positions. In reality, liquidity and institutional mandates may compromise the allocation, particularly when economic trends produce strong signals.

Macro factor hypotheses and empirical tests across sectors

Macro factors and relative energy sector returns

Following conventional views, we hypothesize that energy sector outperformance is more likely in an environment of unusually strong global economic growth, strong local exports, and strong global energy price growth, all of which should support volumes and prices of sales in the energy sector more than on average in the economy. Negative factors for relative energy sector returns are high global inventories and local currency appreciation in real terms, which hurts the local-currency profits of international sales of energy products.

Four of the macro factor candidates displayed significant predictive power for subsequent monthly sectoral return outperformance, more than for any other sector. Excess economic and industry growth also posted a positive forward correlation but with lower probabilities of significance.

The conceptual parity score of the factor candidates has been a significant predictor for monthly relative equity returns over the past 25 years and even over the first and second half of this sample period separately. The Sharpe ratio of the naïve PnL has been 0.6, but with decent seasonality: about 90% of the long-term value was generated in the top 5% of monthly PnLs.

Macro factors and relative materials sector returns

The upfront hypotheses were that materials sector returns would outperform in an environment of strong and improving manufacturing sentiment and high producer price growth. Also, improvement of terms-of-trade in the home country and real effective depreciation of the local currency should support outperformance of materials companies returns. Finally, disparities between refined energy and crude price trends should help part of the materials sector globally.

 Most hypotheses individually failed to produce significant predictive power over the past 25 years. Only the real local-currency depreciation and the refined energy price outperformance yielded high probabilities of a systematic relation.

As a result of the tepid empirical support for most factors, also the conceptual parity score failed to display stable and significant predictive power at a monthly frequency. The relationship with subsequent returns has been positive across sub-samples but modest. Also, judging by the naïve PnL, economic value has been small and extremely concentrated around the time of the great financial crisis.

Macro factors and relative industrials sector returns

Basic hypotheses are that industrial sector returns should outperform in an environment of strong exports and high and improving global manufacturing sentiment. Presumed forces of underperformance are energy price growth (higher costs), real appreciation (lower revenues), and high real interest rates (headwind for capital goods demand).

Unsurprisingly, global manufacturing sentiment stands out as a highly significant predictor of industrial companies’ excess returns over the past 25. Against the initial hypothesis, real currency depreciation has been negatively related to relative industrial returns. The other factors all posted positive predictive relations but with low significance.

The probability of significance of the predictive power of the conceptual parity score of all six factor candidates has been around 85% for the past 25 years. Predictive power for relative sectoral returns has been much softer in the second half of the sample, partly reflecting the “wrong” sign of the hypothesized influence of the real appreciation factor. Value generation has been extremely seasonal, with all of the positive PnL being generated between 2005 and 2012.

Macro factors and relative consumer discretionary sector returns

Consumer discretionary sales are presumed cyclical, typically compounded by inventory dynamics, as in the case of industrials. Most factor candidates aim to capture past excess consumer spending and sentiment for signs of forthcoming setback risk. By contrast, short-term changes in consumer sentiment are hypothesized to bode well for sectoral outperformance. Finally, macro factors that indicate high or rising real interest rates have a presumed negative impact on subsequent relative consumer discretionary returns.

Macro factors of past excess consumer spending and sentiment all have predicted consumer discretionary relative returns negatively and with high significance. However, in contradiction to the related hypothesis, short-term confidence changes have also posted a negative relation to subsequent relative sector returns. Interest rate-related effects were as expected but not highly significant.

Notwithstanding the empirical failure of one of the candidates, the conceptual parity score has posted significant predictive power with respect to subsequent relative consumer discretionary returns over the past 25 years. The predictive relation was also near significance for the first and second sample half separately. Value generation, according to the naïve PnL metrics, has been material, with a long-term Sharpe of 0.5-0.6, but also posted great seasonality, as most macro factors only produce value in times of big cyclical swings.

Macro factors and relative consumer staples sector returns

Most macro factors for the consumer staples excess returns are related to local labor market conditions and retail sales growth. That is because staples’ are presumed to depend disproportionately on lower-income households’ employment and income conditions. Moreover, unlike durable and investment goods, staples sales are expected to outperform relative to other sectors, when real interest rate conditions deteriorate. Finally, high CPI food price growth is expected to benefit consumer staples at the expense of other sectors.

Only excess food inflation posted a highly significant positive relation to subsequent monthly consumer staples returns relative to other sectors. Positive labor market dynamics also predicted sector outperformance positively, but not quite significantly so. Excess wage growth completely failed to comply with the economic hypothesis.

The conceptual parity score posted significant positive predictive power with respect to subsequent monthly relative sector returns for the past 25 years, albeit not for the half samples individually. Long-term value generation, according to the Naïve PnL, has been respectable, with a 25-year Sharpe of 0.6, but extremely seasonal, with all returns being earned in episodes of marked cyclical fluctuations.

Macro factors and relative health care sector returns

The presumed macro factors for health care returns outperformance are local services business survey metrics (of which health care is an important part), as well as local GDP growth and consumer confidence as proxies of health services demand. Meanwhile, indicators of real interest rates and future policy tightening have a presumed positive impact on sectoral relative returns since health care is less sensitive to interest rates than other sectors. Excess core inflation is presumed to be an indicator of the pricing power of local services businesses.

The services sector business confidence score and GDP growth have been nearly significant predictors of relative healthcare returns over the past 25 years. Core inflation completely failed as a positive predictor, while the other candidate factors posted the right sign of empirical relation but struggled with significance.

Altogether, the conceptual risk parity score positively predicted subsequent relative health care returns but failed to do so with high statistical significance. Value generation has been marginal and extremely seasonal.

Macro factors and relative financial sector returns

Factor candidates for financial sector return outperformance focus on a range of indicators of financial health in the economy, namely private sector debt service ratios, government debt ratios, and unemployment rates. Moreover, idiosyncratic factors of profitability of the sector should be credit growth and the real (inflation-adjusted) slope of the yield curve.

Indeed, credit growth proved to be the only highly significant positive predictor of financial sector return outperformance. The credit quality-related indicators all predicted relative returns with the right sign but did not achieve high significance.

Altogether, the conceptual risk parity score predicted subsequent relative returns in the financial sector with high significance for the whole panel and subsamples. Value generation has been fairly consistent across time, except for the period of the great financial crisis.

Macro factors and relative information technology sector returns

Factor candidates are based mainly on the negative role of consumption and broader business cycles: information technology revenues are presumed to be less cyclical due to the low importance of inventory dynamics (except maybe for semiconductors). Information technology is more seen as a growth sector. Moreover, indicators of rising or high real interest rates are presumed to harm technology more than other sectors due to their negative impact on capital spending plans.

Indeed, both manufacturing and services sentiment have been highly significant negative predictors of subsequent month information technology returns versus other sectors, supporting the hypothesis of greater revenue stability. All other factors have a forward correlation with the hypothesized direction but not with high significance across the panel.

The conceptual parity score has been a highly significant predictor of subsequent relative IT sector returns over the whole panel and across sub-samples. Consequently, also value generation according to the Naïve PNL simulation has been material, albeit focused mainly on the 2000s and 2020s.

Macro factors and relative communication services sector returns

The main hypothesis is that (openness-adjusted) real appreciation should be conducive to communication sector outperformance, as it increases local purchasing power while not affecting the competitiveness of communication services much. Other hypotheses are the positive predictive power of services business sentiment, as services include communication and entertainment, the positive predictive power of local employment growth, and the positive predictive power of global core inflation as an indicator of pricing power.

Indeed, only openness-adjusted real appreciation has displayed highly significant predictive power for relative communication services returns. All other macro factors predicted relative returns with the right sign but not with high statistical significance.

The predictive power of the conceptual parity score has been positive and highly significant for the whole panel and sub-samples. Its value generation, according the Naïve PnL simulation, has bee exceptionally strong and consistent, with a long-term Sharpe ratio of over 0.7.

Macro factors and relative utilities sector returns

Presumed key macro factors for utilities are a range of indicators related to local excess production growth and business sentiment. These do not only inform on differences in demand trends across countries and periods, but also indicate setback risk for cyclical industries. As a non-cyclical sector, utilities’ relative returns benefit from past excesses and potential forthcoming inventory dynamics. Another reasonable hypothesis is that utilities’ relative returns benefit from high energy inflation.

All growth and sentiment indicators have displayed positive predictive power for relative utility returns, but only broad GDP growth and services sentiment (which includes utilities’ business confidence itself) have been highly significant over the past 25 years.

In a display of extreme seasonality, the predictive power of the conceptual parity score has been significant only for the early half of the sample. Also, its value generation only lasted from 2000 to 2010.

Macro factors and relative real estate sector returns

Presumed macro factors of real estate sector outperformance include construction and services business confidence, both of which are directly related to the real estate sector. Also, consumer confidence should have a positive effect on demand for real estate, while excess inflation is negative for interest rate conditions and sectoral prospects.

Only short-term changes in services business confidence posted highly significant predictive power for subsequent relative monthly real estate sector returns. All other factors posted the right sign of forward correlation but struggled with significance.

As for the utilities sector, the conceptual parity score’s success of predicting relative real estate returns across the panel has been extremely seasonal. Statistical significance can only be found for the first half of the sample. Value generation has focused only on periods of real estate busts.

A proxy PnL based on cross-sector allocation

The 11 relative sector strategies can be consolidated into a single relative value portfolio that takes long and short positions across sectors without net exposure to the overall equity market. This is a very rough and ready approach. The strategies are not independent, as they draw on a limited set of macro factors that can cumulate in impact upon consolidation. For example, indicators of business cycle risk may drive long positions in non-cyclical sectors and short positions in cyclical sectors.

However, principally, this portfolio is equivalent to an overlay strategy for an index-weighted or equally weighted equity exposure across all 11 sectors. Put simply, the PnL value is additive to a standard equity market exposure. The key question is: has the economic value of such an overlay been material?

The below chart plots all 11 relative sector strategies based on their respective conceptual parity scores. Performances are vastly different in the long run, but all have ultimately been profitable. Also, there value generation across many strategies has been focused on the times of economic crises and recoveries, when cross-market signals were strongest and most unified.

The sectoral relative value strategies for the above chart have all been scaled ex-post to 10% annualized volatility for the purpose of presentation. In practice a risk-parity based combination of these strategies would require sequential absolute or relative volatility targeting, whose implementation depends on many key methodological decisions, which beyond the scope of this post.

As a simple proxy for assessing the value of relative sector allocation, we simply look at an equally weighted composite PnL of the individual (ex-post scaled) sectoral PnLs with monthly re-balancing. The resulting consolidated naïve PnL shows considerable long-term value generation, albeit with pronounced seasonality, as is often the case when strategies are based exclusively on macroeconomic indicators.

The long-term Sharpe ratio from 2000 to 2024 (June) has been 1.3. The Sortino ratio was even close to 2, as downside volatility has historically been limited. The high seasons for the allocation strategies were economic downturns and recoveries. Notwithstanding the pronounced seasonality, the concentration of PnL generation was moderate: the 5% best months of the sample period produced just below 50% of the 25-year PnL.