Fallout from Asian crisis leaves wide range of outcomes possible in outlook for olefins

Walter Sedriks
SRI Consulting
Menlo Park, Calif.

• Projected Ethylene Production Growth [165,841 bytes]
• Ethylene Market Penetration [39,051 bytes]
• Impact of SE Asian Crisi on Ethylene Output [71,189 bytes]
• US Ethylene Capacity Utilization [82,318 bytes]
• Petrochemical Cycle Projection and Impact Scenario [96,458 bytes]
• Ethylene Globalization Continues [70,551 bytes]
• Value-Added Reductions Continuing [72,591 bytes]

The material in this article was developed as part of SRI's Process Economics Program Studies, and some of it was originally presented at the ABB Lummus Global Ethylene Seminar at Napa, Calif., June 1998.

Petrochemical prices have been falling steadily for almost a year, and the financial crises in Asia remain largely unresolved. The outlook for the global olefins and derivatives industry thus appears as uncertain as it has ever been.

Because of the level of uncertainty, a scenario approach helps to provide insights into the directional trends and to assess ranges of uncertainty for the parameters that affect market growth, capacity utilization, and profitability.

Background, methodology

A good perspective on the petrochemical industry is gained by looking at the historical growth patterns for ethylene over the long term.

Because demand forecasting is an uncertain business, examination of an extended time frame is often enlightening. To assess how that trend may change, it is also necessary to examine the disaggregated demand by end product and by region, and to look at its relationships to overall growth of the economy. Doing so typically makes the projections more solid, even though it does not improve predictive ability in any deterministic sense (i.e., simple extrapolation still fails).

Tying growth rates to GDP, for example, provides a good check and reference framework for absolute magnitudes and timing (relation to the overall business cycle). The relationships to GDP are typically considered in terms of "multipliers" and "penetration" curves. These parameters are tied to life-cycle analysis, which anticipates that ultimately growth will decline to GDP growth levels as the product matures.

Multipliers in this context are defined as the ratio of product growth rates to GDP growth rates. They can be useful for orientation purposes, but their use is often confounded by lack of consistency (e.g., different values result from a 5-year running average, a 10-year running average, a trend line). In keeping with life-cycle theory, multipliers will also change radically and unpredictably, as was the case for pre- and post-1970s growth for ethylene.

Another way to look at the relationships to GDP is in terms of penetration levels. As can be seen, however, ambiguity remains: Looking at such plots in the mid-1980s, for example, Exxon Corp. justifiably argued that the ethylene penetration for the U.S. and Europe was already complete and that ethylene demand would grow in these economies only at rates no higher than those for GDP.¹

Note that simple statistical extrapolation on its own can give good predictions over certain periods, even quite extended ones, and has considerable appeal because it has the appearance of avoiding subjectivity. Unfortunately, because extrapolation fails to anticipate turning points, it can also be very dangerous: Good statistical fits for past data can provide an unjustified sense of confidence about forward predictions. Thus, for 1940-70, an 18%/ year exponential growth curve for global ethylene shows a closeness of fit close to unity. A belief in projections for demand growth extrapolated from that well-established trend was a major cause for the excessive overbuilding in the late 1970s and led to the severe restructuring in the industry in the 1980s.

Actual ethylene growth from the early 1970s on dropped to about 5%/year, and the multiplier on GDP growth for ethylene fell from over 6 to less than 2. The likely causes for the sharp fall are a change in the life-cycle phase for ethylene from an initiation to a growth stage in the 1960s, compounded by the oil price shocks of the 1970s.

A burning question currently might be whether the lowered growth seen for the 1988-97 period represents an inflection point or just a period below trend-the difference between the projections based on the 1971-87 and 1988-97 exponential "trend" lines is equivalent to about 30 crackers by 2007.

Thus statistics alone here can provide no definitive answers. The multiplier and penetration curve behavior is helpful but not definitive, because, over the medium term, GDP and ethylene growth are largely decoupled. Examination of the disaggregated regional trends is somewhat more enlightening. For example, the extended and severe disruption to growth in the former Soviet Union clearly contributes to a flattening in the global growth curve. Because of that, in these projections, it is necessary to disaggregate the growth there and make adjustments for the FSU and the rest of the world separately. However, an inflection point may still exist for the latter.

History also tells us little about the likely longer-term consequences of the Asian crises on ethylene growth, projections about which remain largely speculative. Nonetheless, construction of some rational and consistent scenarios can serve to define the likely ranges of uncertainty and indicate the robustness of the scenarios. Scenarios in which a turning point has taken place in the current decade now look quite plausible and should be incorporated in prudent planning exercises.

Cracker capacity additions

To get a good quantitative feel for the range of uncertainty entailed in the demand projections, it is necessary to first translate a selected set of growth projections to an approximate number of additional crackers and expenditures that would be needed over the next 5 years (see table, p. 21).

A 5-year outlook on the supply side corresponds to one where plans to go ahead with cracker construction will typically need to have been made before the end of this year, and on the capacity side can thus be examined in terms of proposals already on the table. In this assessment, 5 years should also provide more than enough time for the effects of the Asian crises to work themselves out.

Presented here are three cases that are straight extrapolations along trend and two that entail adjustments to extrapolated trends. The starting point for the projections is the estimated actual production level in 1997, and the endpoint is the level corresponding to the projected regression trend or adjusted trend for the various cases. The 1997 starting point is close to the trend values on a global basis and above trend for the U.S.

The extrapolations are for regressions for exponential fits for 1971-87 and 1988-97, and the regression is for a linear fit for 1971-97. There are two adjustments in the other scenarios: one for maturing in the industry and one for the impact of the Asian crises.²

In the "As Was" scenario, the adjustment is only for "penetration," i.e., assuming that the industry has reached the state of maturity in which growth starts to approach GDP growth.³

In the "Impact" scenario, the adjustment is both for penetration and for allowing for a fall in growth resulting from the current East Asian financial crises and the weakness that has surfaced in the Japanese economy.4 In this scenario, the Asian crises have a substantial and permanent impact, namely to reduce the absolute demand well below the levels that were previously projected. More specifically, it assumes that both 1998 demand and trend line ethylene demand in 2002 are lower than it would otherwise have been, by the equivalent of about 1 year of global growth. Trend line growth rates after 2002 return to their previously projected levels, but on a lower absolute base, i.e., absolute demand does not return to the previously projected trend line even though growth rates do so. The Impact scenario represents our present base case.

Of the scenarios depicted in the table on p. 21, given a time frame of less than 15 years, the most pessimistic growth case shown, the linear projection, should be heavily discounted. Closer examination shows that the linear correlation implies that current ethylene growth rates are well above trend, and that the ethylene trend growth has already fallen close to GDP trend growth rates. Looking at the ethylene penetration histories (see chart on p. 22), that does not appear to be highly plausible-the apparent flattening of the penetration curves in the 1980s was not sustained, and flattening does not appear likely in the near future for ethylene worldwide. All the other cases in the table, however, represent quite plausible scenarios.

For those scenarios, the range of uncertainty over the next 5 years for cracker building corresponds to about 24 crackers. In our base case assessment, to keep capacity utilization at comparable levels, additional capacity-equivalent to 26 crackers over the period or more than five crackers/complexes/year-is likely to be needed, and in the most optimistic case, double that. The ratio between new grassroots capacity and incremental capacity expansion is expected to be about 2:1. The expenditure on new and expanded cracker capacity is thus likely to be $13 billion or more.

Assuming that, after 2002, the rates of growth return to previously projected trend levels (but on a smaller absolute base) the Impact scenario numbers improve in relative terms. For the 10-year outlook, the range of uncertainty for cracker building corresponds to about 40 crackers. In our base case, additional capacity equivalent to 66 crackers over the decade, or more than six crackers/complexes per year, is likely to be needed, and in the most optimistic case, almost double that. The expenditure on new and expanded cracker capacity is thus likely to be $30 billion or more over the 10-year period.

Regional demand variations

More depth is added by looking at the picture in terms of regional demand growth in the two adjusted cases for the period to 2002.

A good working approximation for the Impact scenario would be simply to proportion the fall in demand across the board in line with historic growth rates. Attempting to proportion out the demand drop in line with a more detailed evaluation of the economic outlook for each region may first appear to be more realistic; but given the level of uncertainty, such disaggregation, when not essential, may just add complexity rather than sophistication. In particular, the extent of global trade in relation to East Asia and the reduction of that trade will tend to spread the decline in growth worldwide.⁵

The results of reducing the trend line demand in 2002 by 1 year's growth across the board, by region in proportion to historic growth rates, are shown in the table on p. 22. In comparison with the As Was scenario-which made adjustments only for penetration-of the overall drop in global ethylene growth of about 5 million metric tons/year, the Asia-Pacific region takes the biggest hit, with annual production that is lower by about 2 million tons, or 40% of the total. In terms of growth rates, the global rate drops by about 1 percentage point, whereas the Asia-Pacific rate falls by about 2 percentage points over the period.

Supply trends, cracker capacity

A noteworthy feature of the petrochemical industry is that supply in the medium term has been highly inelastic. New plants typically take about 3 years to construct (with an even longer gestation period), and even plant expansions take about a year. Also, new capacity requires huge investments and comes in large increments (e.g., a U.S.-scale cracker would currently amount to 3% of U.S. production capacity).

This medium-term inelasticity has been a major contributor to the cyclicality seen in capacity utilization and in profitability. Neither the decrease in the size of the step changes as the industry size increases nor the trend toward reducing lead times is likely to alter the basic picture.

Because of the lead times, reasonably accurate forecasts of capacity additions can be made for the near term (less than 3 years). For longer time frames, however, forecasting the extent and particularly the timing of new capacity and capacity expansions (capacity creep) remains a highly speculative exercise. What appears to be often overlooked or ignored is that feedback has a major influence on capacity additions. The amount added depends not only on demand growth and perceptions about growth, but also on changing perceptions about profitability.

A wild card is the extent of capacity closure. Such closures are rarely publicized, and it is easy to forget how many closures took place in the 1970s and 1980s, when about 22 crackers were permanently closed in the U.S. alone. Such extremes are unlikely to recur, but there are a sizeable number of plants older than 30 years, so it is likely that significant closures will take place during downturns as the global market becomes less restricted and more facilities become privately owned.

Thus even medium-term supply-side projections are more uncertain than they seem at first sight, and are best made on a scenario basis by "iteration" that takes account of the dynamic relationship between the profitability outlook and supply/demand balances.

Given the wild optimism in the petrochemical industry in good times, particularly of the new players, the projections constitute an exercise in discounting, not of cash flows, but of planned cracker numbers.

Under the As Was scenario, of the nearly 70 units approved for construction, about 15 correspond to units that are generally envisioned by the players themselves to start up after 2002. Another 25 are units that are likely to be delayed or never built. The sentiments expressed by some of the protagonists currently suggest that their postponements are a result of the Asian crises, but these are really casualties of the current downturn and the excess capacities proposed.⁶

In the As Was scenario, the largest volume of new capacity added is in the Asia-Pacific arena, while the highest percentage increase is in the Middle East, and as might be expected, the capacity growth pattern reflects that for production growth. In China, the majority of capacity additions in this time frame comprises the expansion of the existing crackers, rather than construction of the proposed major new joint-venture complexes. Substantial capacity is added in the U.S. and Canada, both as newbuilding and as capacity creep. Significant capacity is also added in Western Europe, but mostly through capacity creep. There are only moderate closures of capacity in Europe and the U.S. in this scenario. In Eastern Europe, the growth in demand is met primarily by increases in capacity utilization. In this case, the global capacity growth averages about 4.7%/ year, slightly in excess of demand growth, to give a slight fall in average global capacity utilization to around 83% by 2002.

In the base case, or Impact scenario, a situation of much more substantial postponements is highly plausible, given the current Asian situation.

Under that scenario, the Asian crises continue-and combine with the deepening of the inherent industry down cycle, more sober judgments, or financing inability-to serve to postpone another six major complexes as well as some expansions. Acquisition of East Asian units by some of the global players is implicit, as are more substantial closures of older capacity in both Europe and the U.S. The overall growth in capacity under this scenario averages about 3.5%/year, more than a percentage point lower than in the pre-crisis scenario but still slightly in excess of demand growth, to give a marginal fall in average global capacity utilization, again, to around 83% by 2002.

The two sets of data represent snapshots of highly plausible global supply and demand scenarios for 2002 as viewed before and after the Asian crises. However, they tell relatively little about the outlook in terms of the profitability cycle. On a global basis, the break between the seller's and buyer's markets, at which a cyclical upswing typically takes place, appears to lie at around 85% of nominal global capacity utilization. Utilization levels at 83% suggest a 2002 point that remains marginally in a downturn. If the feedback and retrenchment implicit in the Impact scenario do not take place, and the As Was scenario prevails in the face of lowered demand, capacity growth will plunge utilization to the 70% decile. That would almost certainly result in a more extended downturn and ultimately a much more radical and painful restructuring in the industry.

Historically, better insights on global petrochemical profitability cycle timing can be gained by assessments keyed to data and projections for ethylene in the U.S., rather than on the basis of global analysis, for which data are far more ambiguous.

Timing, profit cycles

The timing of the capacity expansions projected here-and in particular the optimum timing-remains highly moot and will be significantly influenced by what happens in Asia.

In maturing, the basic petrochemicals industry has become commoditized and subject to severe cycling in its profitability. A key determinant of the fortunes of the industry at any given time is the balance between demand and supply. This balance, quantified in terms of capacity utilization, often serves as a measure of people's expectations about product shortages. Those expectations, through the prices buyers are prepared to pay for product, in turn determine profitability. The cycles in capacity utilization and profitability result from a combination of cycling in demand (linked to the business cycle) and the interrelated lagged cycling in capacity expansion (construction cy- cle). Assessing cycle timing is a particularly complex exercise, because it entails quantifying the capacity/demand balance and its likely effect on margins in a situation where substantial feedback takes place.

The approach used here to quantify profitability cycle timing is intended to key projections to ethylene as a surrogate for the industry, and to focus on the U.S. as the global benchmark. The much lower reliability and accuracy of global production and capacity data for ethylene has typically made impractical any approach based on correlations keyed to global capacity utilization.

In both the Impact and As Was scenarios, allowance is made for two new grassroots crackers to be constructed by 2002, and for continuing significant capacity creep (close to an average of 2%/year to 2006). A not insignificant closure of capacity is envisioned in both scenarios, and more so in the Impact case.

The resultant projected capacity utilization patterns for the two non-trend growth scenarios are shown in the chart on p. 23. Upswings typically have taken place when capacity utilization breaches a band around the 90% level. A particularly interesting result is that, for the As Was scenario, this results in a peak in capacity utilization in 2000 and one in 2004-05. In that scenario, the East Asian crisis is only a "blip;" it has no permanent effect on global ethylene demand, which returns to the previous absolute levels by 2000. In contrast, if a "1-year stepdown" in demand is assumed, as in SRIC's Impact scenario, there is not a peak until the later point in time. The projection of a 2004-05 peak thus appears to have some robustness.⁷

A corollary to the projection of an upturn starting around 2003 is that right now may be the optimal time for initiating contracyclical capacity expansion projects. The caveat is that, if the majority of players start to believe in such a scenario for petrochemical profitability cycle timing and are able to go ahead with their expansion plans, the scenario timing then becomes less likely to be played out in practice. The probability of the timing being realized, even in directional terms, is to some extent inversely proportional to the level of consensus attained-realization of the scenario becomes contingent on the majority of players either not believing the scenario, or lacking the resources to act on their beliefs. The paradox arises because of feedback. Thus, to enhance the chances of success, grassroots expansions should typically go hand in hand with a "build-and-scrap" policy that reduces the size of the incremental capacity added yet serves to upgrade the competitiveness of the production facilities.

Margins, prices

The upper chart on p. 24 depicts the Impact scenario projections for U.S. ethylene margins and prices.

Price projections to 2015 are shown in both current and constant dollar terms8 and illustrate to what extent the projection methodology can replicate history when using estimates of actual oil prices and capacity utilization levels. As noted, the U.S. ethylene projections can be considered to represent global benchmarks, and ethylene cycle timing underpins global petrochemical cycle timing. For other products and other regions, if their patterns of supply/demand are anticipated to differ radically from the trends for the benchmark, adjustment in the amplitudes of the price swings needs to be made in line with such expectations. In particular, in the Asian arena, the tendency has been for the swing in prices to be more extreme over the cycle than in the developed areas.

The ethylene price and margin projections are keyed to Arabian light crude oil prices that return to $18/bbl levels by 1999, and thereafter follow a flat oil price scenario, reaching $27/bbl in current dollars (3%/year inflation) or $15/bbl in 1994 dollars by 2015.⁹

Study of petrochemicals production economics over the years strongly suggests that the basic petrochemicals industry will yield attractive returns only to "pacesetter" producers. With that in mind, the price projection scenarios must be subjected to a "reality check" via the lifetime profitability that they imply.

Industry's changing face

The continuing globalization of industry in general and petrochemicals production in particular is unlikely to be arrested. Among the key elements in this are the progressive reduction in tariffs and other impediments to trade. One outcome is that, within another decade, close to half of ethylene production will be located outside the traditional triad of Japan, Europe, and the U.S. (see graph at right in lower module on p. 24).

Within companies, a major factor in enabling true globalization of operations has been the role played by advances in information technology in facilitating the management of complex and dispersed organizations.

The industry remains dynamic and, as shown in the chart on this page, continues to make progress down experience curves, with current cost reductions likely to continue. One interesting result of this progress, for example, has been the reduction in the fixed-cost component for basic petrochemicals and the resulting importance of feedstock costs in the overall equation. Thus, feedstock typically costs comprise over 60% of the cash costs in producing ethylene from ethane. The leverage of feedstock costs has facilitated the entry of new producers with inherently low-cost feedstocks, such as the Saudis, and it also accounts for the competitive edge that can result from optimizing vertical integration with refining.

Despite the importance of feedstocks, given the competitiveness of the global industry and the inherent sophistication of the process technologies, continuing progress down the technology experience curves remains a critical element for success in the industry.

Industry realignments

The three critical factors in this assessment-severe cyclicality in profits, globalization, and continuing and substantial industry experience curve effects-have combined to present a dilemma for industry players.

That has been driving an extensive realignment in the spectrum of participants and the relationships between them. Most of the traditional chemical majors (e.g., Monsanto Co., ICI, Hoechst AG, and, most recently, DuPont Co.) are essentially moving out of basic chemicals production. They have chosen to "upgrade" their businesses into higher value-added areas that were once considered downstream niches related to basic chemicals (such as coatings, pharmaceuticals, and life science products) but have become major, rapidly growing markets in themselves. The trend to go down separate paths has been encouraged by stock markets, which in recent years have tended to reward narrowing of focus and specialization as opposed to diversification-but that could also be subject to a cycle.

Primarily on the basis of the feedstock card, the mantle of leadership of the basic petrochemicals industry in the established markets is finally falling onto the oil companies, who have taken 20 years to realize the potential in petrochemicals they perceived as imminent in the 1970s. Other major global players, again building on feedstock advantage, are the essentially national entities, such as Saudi Arabia's Saudi Basic Industries Corp. and Venezuela's Pequiven SA. The sophisticated process technology and intellectual property rights that remain key elements of the business have led to various forms of joint venture arrangements between established technology owners, large and small, and between them and the newer players that own feedstocks. Other new entrants are the regional players in the developing areas, whose prime inherent advantage lies in the closeness to markets and the avoided transportation costs. The myth of competitive advantages accruing to players with depreciated plant is still sometimes espoused but in reality provides the most assured way to exit from the industry.

The critical mass needed for successful play appears to have increased, probably due to the extent of globalization. Thus continuing amalgamations of moderate-sized players in novel combinations, e.g., along the lines of Equistar (the joint venture of Lyondell Petrochemical Co., Millennium Petrochemicals Inc., and Occidental Chemical Corp.), as well as megamergers among major players may also be anticipated. Competitive forces are likely to maintain the momentum of privatization and move it further upstream.

Successful participation in the industry has therefore become more complex and will need both feedstock and technology advantages, as well as agility, novel partnering approaches, and a genuine concern for the environment.

Growth of the global olefins industry is likely to remain well above global GDP growth levels for the next decade, and sophisticated technology will continue to remain a cornerstone of the industry. For players in basic petrochemicals, the ride ahead, therefore, if not always comfortable, should nonetheless continue to be exciting.

References

1. See K. N. Robertson, VP Basic Chemicals, World Olefins Outlook, World Chemical Congress of Marketing and Business Research, Newport Beach, Calif., Sept. 10, 1986.
   
2. It is also possible to envision a number of plausible and radically different scenarios, mostly environmentally driven, in which, for example, a switch to natural gas-based fuels commences. However, this would not have a major effect on petrochemicals within the time frame of concern here and would probably affect refining much more.
   
3. Note that projections for ethylene made along similar lines more than 5 years ago worked well on an aggregated global basis, but underestimated ethylene production growth in the U.S.-see SRIC's Process Economics Program (PEP) Report 205, W. Sedriks, Petrochemical Industry Profitability, July 1993.
   
4. The reference to the Southeast Asian crises in some of the figures and tables should be taken in this broader context.
   
5. For example, about 5% of U.S. ethylene production in recent years has been exported as derivatives directly to East Asia, an amount close to 2 years' growth in projected U.S. ethylene production.
   
6. The base case scenario over the past decade has entailed a downturn commencing in 1998 and always has been somewhat conservative in severely discounting the new capacity announcements. For example, following the 1994-95 profitability fly-up, SRIC projected that only about half of the more than 60 announced crackers to 2000 would be built-see W. Sedriks, Ethylene Profitability Outlook, Schroder-Wertheim/Chemical Week Annual Chemical Conference, New York, Dec. 4-5, 1995.
   
7. There may also be some inbuilt bias here, because that timing matches the base case used for the last 4 years. Interestingly, the earlier base case timing was keyed to the assumption that the type of longer-term cycle for petrochemical profitability shown over the last 2 decades would continue (a cycle with a peak-to-peak time of about 8 years), with a downturn, driven by excess capacity, developing in 1998. Given the extraordinary strength of the U.S. economy, it is now debatable whether a capacity-driven downturn would necessarily have materialized without the Asian financial crises. The fall in demand due to the Asian situation does appear to have ensured that such a cycle will indeed continue for at least one more swing, and that its severity is increased. The Asian situation itself can also be viewed as a manifestation of inherent cyclicality in the global economy.
   
8. Assuming two similar cycles in tandem.
   
9. See PEP Review 94-3-4, or W. Sedriks, Projecting Crude Oil Prices, Chemtech, February 1998, pp. 47-53.
   

The Author