NYU Stern School of Business | Full-time MBA, Part-time ...



Goldman Sachs

Strategy Focus

Global Portfolio Strategy

Local meets global. Everyone is global these days, but some regions are more global than others. The US, UK and European equity markets are predominantly global markets. But the emerging markets especially, as well as Japan, have strong local factors.

Global sectors, local sectors

Technology, energy and telecoms are the most global sectors. Industrials and materials are the least global sectors. Stock-specific risk also varies across sectors: energy and utilities have the lowest stock-specific risk, while consumer cyclicals, technology and industrials have the highest.

Earnings are global

Global sector effects dominate local or country influences on earnings revisions. The global earnings effect is most marked for Europe and the US. The UK has the smallest local element to earnings revisions. Technology, energy and telecoms have the largest global component to their earnings revisions; utilities has the lowest global sector impact on earnings revisions.

Locals and globals fight for the upper hand

The strength of local and global effects varies over time. Global sector effects are strongest when regional economies are in sync. Local effects are more important when regional economies diverge.

Global ascendancy

We believe the global recovery will be a synchronised one, so global sector effects are likely to remain important.

Asset allocation cross-dressing

Relative performance across regions is a combination of local and global effects. Global effects arising from differences in sector weightings dominate pure local effects in driving relative returns across regional indices.

Strategy Focus Global Portfolio Strategy

Executive summary: Local and global influences on stock returns

Why they happen ...

• The relative importance of local, global and stock-specific influences on equity returns varies significantly across the major markets. The emerging markets and Japan have significant country risk factors. By contrast, relatively little of the variation in US and UK equity returns reflects local factors. Europe (ex-UK) fits somewhere in between.

• Sectors also differ in the sources of their returns. Technology, energy and telecoms are the most global sectors. Industrials and materials have the lowest global factors.

• Stock-specific risk varies a great deal across sectors. Energy and utilities have the lowest component of stock-specific risk. Because they are the most homogenous sectors, there is probably less scope for active stock selection. The largest stock specific risk occurs in consumer cyclicals and technology. There may be more scope for active stock selection strategies within these sectors, although these opportunities are available only because of greater risk.

• The fundamental reasons for a stronger global effect include a significant "global" element to sector earnings revisions. Changes to sectoral earning estimates are closely correlated across regions. In addition, as regional interest rates usually move in the same direction, there is limited scope for divergent valuations across regions.

• Not surprisingly, the emerging markets and Japan have the highest source of local variation in their earnings revisions. By contrast, the global sector influence on European, US and UK earnings swamps the local effect.

• Technology, energy and telecoms have the largest "global" components to their earnings revisions, and utilities the smallest. The sectors with the largest amount of stock-specific earnings risk are materials and consumer cyclicals; the sectors with the lowest individual earnings risk are utilities and energy.

... and what they mean for investment strategies

• Many asset allocation strategies are designed to meet future liabilities by maintaining a large proportion of assets in the equivalent "local" equity market. These strategies might be reasonable in the long run (especially if regional growth rates are becoming more coordinated) but there must be significant scope for short run deviations of the growth rate of "local" assets from the growth of liabilities, given the global nature of today's equity markets.

• Sector and local influences on stock returns vary through time. The importance of global sector effects tends to increase at the expense of country effects when economic growth rates across the world are highly synchronised. The same applies when interest rate synchronisation increases.

Global Portfolio Strategy Strategy Focus

• The global recovery, when it arrives, is likely to be synchronised. We believe that global sector effects are likely to remain in the ascendancy over country effects.

• For asset allocation strategies, differences in performance between regional indices reflect a combination of sector and country effects. Sector effects arise from differences in sector exposures across regions. Country factors reflect differences in the relative performance of the same sectors across regions.

• Japan is biased towards consumer cyclicals and industrials while the emerging markets are biased towards energy, materials, technology and telecoms. The UK is the most defensive market in its sector make-up, as it is concentrated in consumer staples, energy, financials and healthcare. The US and Europe fall in between these two extremes.

• Understanding the role of local versus global factors on index performance, identifying when these factors are likely to strengthen and weaken, and recognising the feedback from sector performance to country performance should be an essential part of consistent global portfolio construction and risk control.

Strategy Focus Global Portfolio Strategy

Part One: Some worthy global and local attributes

Revisiting the sector versus country debate

The increasing importance of global sector influences on individual equity returns is generally well accepted (and extensively researched; see "Sun, Sand and Sectors", August 2, 2000). Exhibit 1 updates the contribution of global market, global sector, local and stock-specific influences to world large-capitalisation returns. Exhibit 2 shows the same graph for all stocks in the MSCI developed countries universe.

[pic]

The strength of global (global sector and global market), regional and stock-specific factors varies from one index to another. Exhibit 3 shows that stocks domiciled in Japan and the emerging markets are much more sensitive to local risk than stocks in the US, UK or Europe.

There are also differences in local versus global influences in returns across sectors. Exhibit 4 shows that the three "most global" sectors are telecoms, technology and energy. The "least global" sectors are industrials and materials, while all the other sectors have a similar global influence. We have to be careful interpreting these results. After all, it is difficult to believe that industrials and materials are "less global" than utilities. Our framework simply cannot explain as much of the variation in stocks in these two sectors as elsewhere. Maybe the sectors are not homogeneous. Note that the stock-specific component of returns in both industrials and materials is higher than average, so we are unable to explain much of the variation in returns with our model.

[pic]

Strategy Focus Global Portfolio Strategy.

The fundamentals behind rising global influences

Stock returns reflect changes in expectations about earnings growth (ie earnings revisions) and/or changes in the value attached to earnings. If earnings revisions are more global than local, then stock returns might logically be expected to be more global than local. Maybe company profitability has become more linked to the fortunes of the global business sector in which companies operate than to the country in which the firm does most of its business.

Overall, we find that earnings revisions are much more heavily influenced by global factors than by local factors (see Exhibit 5).

[pic]

There are significant variations in the strength of the global and local influences on earnings revisions across the major markets. We also find differences in earnings influences across sectors.

At the regional level we find that emerging market and Japanese earnings revisions (see Exhibit 6) are more heavily influenced by local factors than UK (the lowest local effect), US and European earnings revisions. This seems consistent with the larger local influences we see on Japanese and emerging market equity returns.

At the sector level, technology and energy have the "most global" component to earnings revisions, followed by telecoms. Utilities is the "least global" (see Exhibit 7). Exhibits SI-S4 at the back of this report illustrate these global earnings links using European and US sector earnings.

[pic]

The relative importance of local versus global effects on earnings and performance changes through time. The global influence on performance has trended higher (see Exhibits 1 and 2), but there is also a cycle. The fluctuations in the relative importance of the global sector and local influences reflect the business cycle. The global sector effect becomes more important, and the country influence less important, as regional growth rates converge. For example, the higher the degree of synchronisation of regional business cycles, the higher the synchronisation of regional profit growth rates. When regional growth rates are synchronised, there is much more scope for sectoral growth rates to differentiate stocks' performance. Exhibit 8 shows that the average country effect varies according to the variability of regional industrial production growth rates.

[pic]

Global Portfolio Strategy Strategy Focus

Part Two: Time to visit the relatives!

Performance is all relative

Relative return differentials between regional indices reflect the interplay between global sector and local market influences on the regions. Consider the relative performance between the Japan index and the US index.

R j = Global + Global Sector + Country j + Stock Specific j

R US = Global + Global Sector + Country US, + Stock Specific US

R j - US = Global Sector j- US + Country j- US

The difference between Japanese and US performance boils down to differences in sector and country effects. The global influence is the same for each region (by definition) so it cancels out. The stock-specific influences sum to zero in each region, so they also cancel out.

Here is an example of how to identify the global sector effect on relative index performance between Japan and the US.

Exhibit 9: The sector effect – A simple example

US Japan Global sector

Technology 6%(75) 6%(50) 6%

Consumer staples 2%(25) 2%(50) 2%

Country 5% 4%

Source: Goldman Sachs Research.

There are two regions and two sectors. The Japanese index comprises 50% technology companies and 50% consumer staple companies, all the same size. The US index comprises 75% technology companies and only 25% consumer staples, all the same size. Note that the performances of the two sectors are identical across the regions - a 6% return in technology and 2% in staples. However, the US index has produced a better return (5%) than the Japanese index (4%). The US outperformance is solely attributable to the different weightings of the technology and staples sectors in the two indices. There is no differentiation in performance on a sector-by-sector basis. The positive sector effect for the US reflects the higher weighting of the US index towards the best-performing sector - technology.

The next example shows how a country effect operates. We have made a couple of changes to the first example. First, the sector weights between technology and consumer staples are 50:50 in both countries. Second, the returns to the technology sector are no longer the same in both regions. Japanese technology returned 4%, while the US returned 6%. The returns to staples are the same in both regions at 2%.

Strategy Focus Global Portfolio Strategy

Exhibit 10: The country effect - A simple example

US Japan Global sector

Technology 6%(50) 4%(50) 5%

Consumer staples 2%(50) 2%(50) 2%

Country 4% 3%

Source: Goldman Sachs Research.

In this example, the US index has again produced a better return than the Japan index. All of the outperformance is attributable to a country effect. The country effect reflects the difference in performance between regions on a sector-by-sector basis. US technology stocks have performed better than Japanese technology stocks. US staples have performed in line with Japanese staples. So the net effect is a positive country effect for the US. There is no sector effect operating in Exhibit 10 because the sector weights are identical across the regions.

There is a common thread between how we think about sector and country effects for relative performance attribution in the above examples and how we thought about them for absolute performance attribution in Part One. For absolute performance attribution (ie what drives the returns to a specific index), the more closely sectors move together across the world, the stronger the sector influence on index returns. Similarly, when we look at relative performance attribution, if sectors are performing very closely in line with each other across regions (as in Exhibit 9), the main distinction between regional index returns comes down to which region has the dominant share of the best-performing sectors. This is the sector influence on relative index returns.

A rather weighty matter: The sector effect

Sector effects are simply the difference in sector weightings across regions. if a region has a relatively high weight in a sector that is outperforming the rest of the world, then that region will enjoy a positive sector effect. It is easy to identify the indices that will do relatively well (enjoy a positive sector effect) when different sectors are enjoying the best performance. Exhibit 11 shows sector weights for the major regions.

Exhibit 11: Sector weights, region by region

us Eur ex UK UK Japan Emerging

Consumer cyclicals 15% 11% 11% 28% 6%

Consumer staples 6% 8% 8% 4% 7%

Energy 5% 9% 12% 1% 10%

Financials 17% 27% 26% 17% 16%

Healthcare 15% 10% 17% 6% 2%

Industrials 11% 8% 5% 17% 7%

Materials 3% 5% 3% 6% 15%

Technology 18% 8% 1% 13% 15%

Telecoms 7% 9% 13% 4% 18%

Utilities 4% 5% 3% 4% 5%

Source: Goldman Sachs Research as at the end of july 2001.

Global Portfolio Strategy Strategy Focus

Counting the country effect

The country effect can be thought of as the "sector-adjusted" relative return between two regions: it "strips out" the impact of differential sector weights on regional performance. The simplest way to identify the country effect is to look at the world on a sector-by-sector basis. The stocks that make up each sector are domiciled across the major regions as shown in Exhibit 11. For each sector we calculate the "beta" of one region relative to the entire global sector. Usually when we think of beta we calculate the beta of the individual sectors on a region-by-region basis - down the columns in Exhibit 11. This time we go across the rows, sector by sector. The betas are essential to understanding the country effect. They give us some guidance about which region's stocks are most likely to outperform when a particular sector is performing (or not performing). Exhibit 12 provides some estimates of sector betas based on sector performance over the last five years.

Exhibit 12: Sector betas, sector by sector

us Eur ex UK UK Japan Emerging

Consumer cyclicals 1.12 0.86 0.62 0.94 0.90

Consumer staples 1.09 0.86 0.94 0.60 0.89

Energy 0.96 1.08 0.88 0.64 1.41

Financials 0.98 0.90 0.86 1.20 0.91

Healthcare 1.25 0.60 0.75 0.51 0.59

Industrials 1.03 0.86 0.89 1.03 1.29

Materials 1.14 0.93 0.98 0.81 1.20

Technology 1.02 0.92 1.13 0.93 1.24

Telecoms 0.67 1.13 0.78 n/a 0.97

Utilities 1.07 1.08 0.71 0.70 1.48

Source: Goldman Sachs Research.

Exhibit 12 suggests, for example, that if consumer cyclicals are expected to outperform the global market, US consumer cyclicals probably offer the best returns within the consumer cyclicals sector.

We saw in Part One that the importance of country effects varies both across regions and through time (Exhibits 3 and 8). The country effect is relatively powerful in Japan and the emerging markets compared with the US, Europe ex-UK and the UK.

The strength of the country effects in Japan and the emerging markets reflects the wide range of betas shown in Exhibit 22. Wide variations in betas are simply another way of saying that there are larger than average country effects in operation.

Country effects are at their strongest when regional growth rates diverge. When a region is enjoying strong growth relative to the rest of the world, that region's country factor is generally positive. After all, most sectors perform better with stronger growth than with weaker growth, and this holds when comparing stocks within the same sector. To illustrate this effect we show the country factor for Japan versus the difference in Japanese and global growth in Exhibit 13.

[pic]

However, the performance of country factors through the business cycle will also depend on where the sector betas are the strongest. For example, the US and Europe both have higher than average sector betas for utilities. Utilities tend to enjoy their best relative performance when economies are weak, which helps explains why the US tends to have a positive country effect in periods of economic weakness (see Exhibit 14 - all these country effects are presented in Exhibits S5-S8 in the appendix to this report).

[pic]

Global Portfolio Strategy Strategy Focus

Combining sectors and countries: Implications for strategy

The total returns from one regional index relative to another reflect the combination of global sector and country effects. Combining the sector weights in Exhibit 11 and the country effects in Exhibit 12 allows us to:

1. understand how sector and country performance interact;

2. test the consistency of sector and asset allocation views;

3. cross-check whether views on the global and regional economic cycles fit in not just with sector views but also with asset allocation views.

Combining the sector weights (Exhibit 11) and the sector betas (Exhibit 12) yields Exhibit 15 - the impact on relative index returns when any individual sector is outperforming (or underperforming). Exhibit 15 shows the sector-weighted and betaadjusted proportional returns to the major regions assuming a 1% excess return to each global sector, holding returns to all other sectors at zero. For example, a 1% excess return to global consumer cyclicals (assuming all other sectors have zero excess return) will generate the strongest excess return to the Japanese index followed by the US index. The UK index provides the lowest excess return when global consumer cyclicals are the best-performing sector. We have circled the best performing regions (columns) along each row (sector).

[pic]

The key points are as follows.

• Sector (or weighting) effects dominate country effects in determining the best regions to be in for any given view on a sector.

• Japan is the best-performing index when consumer cyclicals and industrials are outperforming. That is why Japan is a cyclical market. An overweight asset allocation position towards Japanese (and emerging market) equities is a positive view on a global cyclical pick-up.

Strategy Focus Global Portfolio Strategy

• The UK is the best-performing index when consumer staples, energy, financials and healthcare are outperforming. That is why the UK is a defensive market. An overweight position on UK equities is appropriate for an asset allocator with a below-consensus view on growth.

• The European (ex-UK) market is slightly more defensive than cyclical. The European index provides the best relative performance when consumer staples, financials, telecoms and utilities are the best-performing sectors.

• The US index is mixed in terms of economic exposure. On the one hand, US relative performance tends to be best when consumer cyclicals, health care, industrials and technology are the best-performing sectors. On the other, the US also performs best when healthcare is in the ascendancy.

The US has experienced the strongest positive country effect recently. And the 11 mixed" sector exposure noted above means that the overall US market has outperformed, as usual, during the current economic downturn. The UK market, as expected, has held up reasonably well, not so much because of a strong country effect as because of its defensive sector bias. On the other side of the equation, Japan has suffered a negative country effect and its sectoral bias towards economic sensitivity has dragged the market down.

Going forward, we expect the global economy to recover. Our framework sends the strongest positive signals on Japan and the emerging markets (where we are overweight) and the strongest negative signals on the UK (where we are underweight). While the US country effect is likely to go into reverse, we are cautious about underweighting the US market too aggressively because the US does not have a clear sectoral bias towards either recovery or downturn. We have a small underweight stance towards the US. Our models do not send a particularly strong signal on European equities. The European country factor is currently in negative territory and it should recover. However, the sectoral bias of Europe is not particularly attractive during an economic upturn. On balance, we are neutral on Europe.

Neil Williams

Alain Kerneis

August 17, 2001

APPENDIX

[pic]

[pic]

[pic]

Decomposition of stock returns

We estimate the decomposition of stock returns or three-month earnings revisions in two steps: first, estimation of regression factors and, second, estimation of their relative importance.

Step 1: Estimating the regression coefficients

The following regression model is run each month on the 2,083 constituents of the MSCI world all countries index.

Ri,t = α t + βcountry i,t • Country i + βsec tor i,t • Sector i + εi,t

with the regression variables:

Ri,t monthly price change of equity i in local currency

Country i dummy variable based on the country of origin of equity i

Sector i dummy variable based on the FTSE world index sector classification

The regression provides an estimation of the following four components:

α t global market (fraction of returns common to all stocks)

βcountry i,t local market (returns explained by countries)

βsec tor i,t global sector (returns explained by countries)

εi,t stock-specific component (regression residuals)

Step 2: Estimating the contribution of each of the four components

We then calculate the contribution of each of the four components to stock returns each month t.

Component Contribution

Global market n • α 2t / Total square

n

Local market Σ (βcountry i,t)2 / Total square

i =1

n

Global sector Σ (βsector i,t)2 / Total square

i =1

n

Stock specific Σ (εi,t)2 / Total square

i =1

n n n

Total square n • α 2t + Σ (βcountry i,t)2 + Σ (βsector i,t)2 + Σ (εi,t)2

i =1 i =1 i =1

[pic]

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download