API Manufacturing: Facts and Fiction

Source: http://www.contractpharma.com/issues/2012-01/view_features/api-manufacturing-facts-and-fiction

The text that follows is owned by the site above referred.

Here is only a small part of the article, for more please follow the link.

By Peter Pollak, Andrew Badrot and Rolf Dach | January 23, 2012

Have costs of Chinese & Indian fine chemical producers closed in on European & U.S. Levels?

In the context of manufacturing, globalization has been considered a one-way street, namely offshoring from industrialized to developing countries. However, as the business landscape of major Low Labor Cost Countries (LLCCs) gradually approaches that of High Labor Cost Countries (HLCCs), economists are increasingly questioning this formula. In a recent publication, the Boston Consulting Group (BCG)1 noted, “China’s overwhelming product cost advantage over the U.S. is shrinking fast. Within five years [. . .] rising Chinese wages, higher U.S. productivity, a weaker dollar, and other factors will virtually close the cost gap [. . .] for many goods.” Later in the article, BCG qualified its statement: “For many products that have a high labor content and are destined for Asian markets, manufacturing in China will remain the best choice because of technological leadership or economies of scale.”
A closer look at the underlying math shows that this process will take much longer than many executives from HLCCs anticipate . . . and would like to see happening.
Executives of western pharmaceutical and fine chemical/custom manufacturing (CMO) companies, too, have expressed the view that the escalating labor costs of the Asian (especially Chinese and Indian) fine chemical companies bring their product costs up to Western levels and impair their competitive advantage. As an additional argument pro domo, quality concerns, which play a particularly critical role in this business, are mentioned.
The global strategies, however, are not congruent, and companies continue to invest in Asia.

The intent of our analysis is to provide objective insights into the competitive landscape between the top-tier fine chemical companies in LLCCs China and India on the one hand, and their counterparts in HLCCs Europe and the U.S. on the other hand. According to Thompson Reuters, 22 fine chemical companies out of 165 in India and 13 out of 214 in China qualify as “top tier”.4 In order to present a complete picture, we consider both hard facts and soft issues. The former comprise three pivotal cost elements: 1) labor, 2) investment and 3) standard operating cost, while the latter tackles the Total Cost of Ownership (TCO). Because of their unpredictability, the impact of exchange rates is disregarded.
Labor Cost
Apart from raw materials, labor costs are the determining element of the “Standard Operating Costs” (SOCO), which will be used as THE main benchmark for our assessment. Labor costs have a twofold effect on SOCO, impacting both fixed costs (which also include investment costs) as well as variable costs (which include conversion costs). In comparison with western (HLCC) countries, particularly the U.S., home of the largest number of “Big Pharma” companies and therefore the largest API market, outsourcing production to Asian countries is mainly driven by their large labor cost advantage. Even by taking the lower labor productivity of the latter into consideration, they still benefit from a three to four times advantage (see Table 1).

The calculation of the number of years it will take until the Chindian labor costs will have closed in on those in the U.S. is based on the present labor costs for chemical operators. Both in China and India, they vary depending on the location. Representative wages are $6,000 a year in India and $4,500 a year in China. In the U.S. the average was $55,000 in 2010 according to the U.S. Bureau of Labor statistics (see Table 1). For the U.S., a compound annual growth rate (CAGR) of 3% has been used. For Chindia, two levels have been assumed: 10% and 15%. The 15% rate takes into account the impact of the ever-growing senior manager tranche in the workforce, especially “returnees” who have completed post-doc studies and worked in western countries. They command higher salaries and this inflates the average salary across the labor pool. For instance, the salary of an Indian Ph.D. with a U.S. postdoc averages $27,000 per year, nearly twice that of a Ph.D. without overseas studies. The result of the extrapolations is shown in Figure 1. The earliest break-even point will take place in 2030 for the “India, CAGR 15%” and the latest in 2050 for the “China, CAGR 10%” growth modes, respectively. This is far beyond any strategic horizon!

Nonetheless, as the BCG study highlights, the pure cost of labor is not enough of a metric to determine the flow of outsourcing/insourcing. Productivity per employee has to be taken into account as well. As shown in Table 1, it is three times lower in India than in the U.S. Taking this into consideration, the overall salary mass per $ output would already be on par in 2020 (instead of 2030) in the “India, 15% CAGR” case. Simulating such an outcome is rather complex, as productivity in Chindia is also gradually increasing along the wage increase. The two metrics are highly correlated. This new reality can be corroborated through many discussions

with local Chindian manufacturers who highlight the fact that they, too, have now to deal with considerations of staffing limitations and optimization. These considerations were irrelevant in the past, when throwing bodies at the problem was their modus operandi.
In absolute terms, typical conversion costs are $50/kg in the west, $23/kg in India and $18/kg in China, respectively.
Investment Cost
Investment costs6 for multipurpose fine chemical plants vary considerably, depending on the location, available infrastructure, size of vessels and degree of sophistication (e.g. automation, quality of equipment, high end or only basic cGMP standards, degree of containment). In terms of investment per m3 reactor volume, Roche’s Process Development and Bulk Manufacturing Plant in Florence, SC, U.S. is generally considered the most expensive plant in the world, with a total investment cost in excess of $600 million. Pegged against the total reactor volume of 50 m3, this translates into $12 million/m3. In comparison, state-of-the-art cGMP multipurpose plants built by a western fine chemical company in the western hemisphere, equipped with 3-6 m3 reaction vessels with a total volume of 100 m3, plus an appropriate number of centrifuges and dryers, would cost approximately $100 million, or $1 million/m3 reactor volume. The most recent, highest standard Indian fine chemical plants are not far-off. Figure 3 shows the pharmaceutical fine chemical plant under construction by Hikal Ltd. at its Bangalore site.

A comparison of investment costs of mid-sized cGMP multipurpose plants is shown in Table 2. The 400-fold difference between the lower cost 1 m3 volume ($30,000 per m3) and the highest unit costs ($12 million per m3) is staggering! The main differentiator is the location in a HLCC, as opposed to a LLCC, as clearly evidenced by the table. Further cost determining factors are the type of industry (pharmaceutical or fine chemical) and degree of sophistication. In the case of plants built in Chindia, it also makes a difference whether the project is managed by local or western engineers. An example in case is the Esteve Quimica’s main production site in China: the investment for the plant, which became fully operational in 2011, was $54 million and has a total reactor volume of 160 m3. This corresponds to a specific investment of $335,000 per m3 and thus the lower end of the range for western pharma fine chemical plants.


  1. Self-sufficient site, comprising admin building and labs, highest standards
  2. High containment bay in fine chemical complex, designed for HAPI production
  3. Self-sufficient site, originally built by Warner-Lambert for Lipitor, later upgraded by Pfizer. Assets include a 6 m3 spray-dryer
  4. State-of-the-art cGMP plant, 4-6 m3 reaction vessels, excluding building and utilities
  5. Fully automated, highest standards, supports & utilities included; built in 2000
  6. First bay of state-of-the-art, five-floor API plant. Total reactor volume after completion: 360 m3. Agitated nutsche filters In-process quality control.
  7. Not fully cGMP plant, 1-5 m3 reactor vessels, including building and utilities
  8. State-of-the-art API plant
  9. Good GMP China standard with full EHS; dedicated production line

Based on a larger cohort of plants, the following ranges of specific investment costs apply (see Figure 3).

In addition to lower costs, investments in new plants in Chindia are also attractive, because construction times are shorter. Building permits and operating approvals are obtained quickly and the abundance of construction workers accelerates the construction itself. This is particularly the case in Special Economic Zones (SEZs), where an ample infrastructure already is in place. Contrary to common belief, high cost of capital in India does not have a negative impact. Indian API exporters also benefit from “External Commercial Borrowing (ECB),” which grants low-interest loans. Additional incentives are granted in China and India for investors in SEZs.
Last but not least, low investment costs result also in a more favorable ratio of net operating assets to sales. Whereas it typically is about $1 NOA per $1 sales in HLCCs, it only amounts to $0.8 per $1 sales in LLCCs.
Standard Operating Cost
Standard Operating Costs (SOCO) are the preeminent measure for cost competitiveness of fine chemical companies. Raw materials aside, labor costs are a major constituent of the SOCO. The criterion plays a dominant role since the very early beginnings of process development. An assessment of SOCO in LLCC and HLCC is used to substantiate the core thesis of this article. SOCO include: utilities, direct labor, depreciation, maintenance, QA/QC and EHS (Environment, Health & Safety). SOCO are determined by the volume time efficiency of the performed reaction. Raw materials costs, which are to a large extent the same across geographies, and Sales, General & Administration (SG&A) costs, which are company specific are excluded.
A precise calculation of SOCO is a demanding task. On the basis of lab-scale and sometimes even pilot plant experiments, one may not accurately forecast which cycle time will be realistic on production scale later-on. Multipurpose plants are used for multistep chemical reactions with widely different throughputs. The products, mainly isolated as dried solids, are manufactured in campaigns occupying the equipment to different extents. Also preparing the equipment for a chemical reaction or cleaning it afterward blocks production capacity and incurs changeover costs. Moreover, costs elements such as labor, capital, utilities, maintenance, QA/QC, and EHS cannot be allocated unambiguously.
A pragmatic approach for determining the conversion cost consists in calculating the volume and time-specific standard operating cost of a plant. Standard operating costs are defined as follows6:

The SOCO calculation is an effective tool for the plant manager to build an overview about cost and capacity. Nevertheless, a weak point in a SOCO calculation is that liquids, which do not require centrifuges or dryers, can show an attractive profit margin without providing a good return for the overall investment in the multipurpose plant.
The time required for carrying out a chemical reaction in a multipurpose plant has both a fixed and a variable component:


  • The duration of a chemical reaction is determined by its kinetics and is therefore a fixed parameter.
  • The time required for raw material charging, heating and cooling are also more or less fixed parameters.
  • The changeover time between single product runs and, more so, production campaigns (clean-up, repairs, plant adaptations, test runs, etc.) depends on the overall operating efficiency at a given site. It has a considerable potential for optimization.

An example for the investment cost for a 60 m3 multipurpose plant in China is given in Table 3.1. The corresponding yearly costs are shown in Table 3.2.


The resulting standard operating costs are:

When comparing SOCO between east and west, it is important to assume the SOCO of a factory at a specific capacity utilization. It is reasonable to assume for western CMOs an average plant utilization rate above 60%, while Asian sites run at an average of 30 – 50%. Actually, the quality of the equipment and installation are often lower than in western CMOs. Also, all reactors of older Asian plants are installed on just one floor. This increases changeover time and therefore decreases capacity utilization.
In Table 4, SOCO for three widely used reactor sizes are listed. The layout of the installation and piping are nearly the same.

In China and India, standard reactor sizes are 2 and 3 m3. Vessels of this size are available at very attractive prices. Sometimes, they even are assembled on-site. Installation costs are also very advantageous compared to a large-scale setup based on 10m3 vessels, as frequently installed in western fine chemical plants. As shown in Table 4, SOCO, despite lower capacity utilization are two to three times lower in Chindia than in the west.
Capacity Utilization
Capacity utilization is one of the key elements impacting costs for CMOs, often determining whether a business is profitable or not. A reduction from 60% to 50% utilization results in a 20% increase of the SOCO (see Figure 4); it will generally wipe out the profit margin. This is all the more important as, under the prevailing competitive climate, it is unrealistic for a CMO to expect a customer to compensate for idle capacity, even if it results from last-minute revised forecasts or even last-minute order cancellations. Internal measures for improving capacity utilization include:

  • Developing new business
  • In-sourcing products that have been outsourced
  • Asset stripping


Total Cost of Ownership  
When someone buys a loaf of bread at a Baker y near home at a price of $2.50 (“ex works” price, in industrial terminology) one would not consider other costs that may be accounted for, such as the time needed to walk to the bakery, the possibility of dropping the loaf of bread on the sidewalk and having to buy a new one, etc. The total cost of purchasing the loaf is simply its purchase price: $2.50.
The situation is different for APIs. If a life science company outsources the manufacture of a fine chemical, additional costs have to be considered on top of the invoiced product price. They comprise internal costs, such as those related with technology transfer, travel expenses in connection with visits to the supplier’s premises, costs for legal advice, auditing and qualifying a new supplier and registering the product (see Figure 5). Furthermore, abnormal situations, such as delayed or incomplete deliveries, resolving quality problems and IP violations, have to be taken in due consideration. They are likely to occur more often in Asia. In addition, running a local procurement office in Asia adds to the required fixed cost structure.
For a meaningful comparison, it is mandatory to determine the Total Cost of Ownership (TCO). This concept has been developed in the 1990’s and has been widely adopted by the pharmaceutical industry7. Thus, the total cost of ownership of the purchase can be much higher than just the invoiced price and make the outsourcing option in general, and offshoring to an Asian supplier in particular, unattractive. With regard to transportation costs, the long distance between customer and supplier is partially counterbalanced by the proximity of the raw material sources. In fact, many raw materials and basic intermediates, for example, the dyestuff family tree, are only produced in Asia. Typically, the smaller the size of the deal and the more demanding the tech transfer is, the less attractive the TCO becomes. There are cases where costs of several million dollars accrue for a complex outsourcing project. Outsourcing to an Asian fine chemical company is also disadvantageous in the event of urgent deliveries, or supply of an exclusive product as compared with a standard one, where IP issues and dependency on one supplier are not at stake. The issue boils down to a trade-off between a $30 million revenue loss, if, say, a $1 billion per year blockbuster drug runs out of stock for 10 days because of delayed receipt of the API against a few percentage points lower unit cost. The impact of TCO is mitigated in the case of an extended customer / supplier relationships, where the costs can be distributed by continuing product orders.
Having an accurate view of the Cost of Quality in pharma is not straightforward as one may think. Taking into account the cost of quality assurance (QA) and quality control (QC) personnel, the costs of batch rejection and ancillary costs lowballs the figure drastically. Taking a process from a 4∑ (with 99.4% accuracy) to 6∑ (with 99.99966% accuracy) can radically increase costs.

And the example to illustrate the point is quite simple: Lipitor, the world’s best selling drug with peak sales of nearly $12 billion in sales per annum. A loss of one day of Lipitor sales worldwide would have impacted Pfizer to the tune of $33 million in those days. Pfizer therefore has multiple suppliers, factories spread across various continents, inventories across regions, etc. All those are part of the Cost of Quality, or otherwise stated as the cost of uninterrupted supply.
By various measures, India provides the world with a significant share of APIs and final formulated drugs. Indian contract manufacturers have experience, but despite this experience, quality problems still arise there more frequently than with western suppliers.
There have been four significant instances of quality issues at major Indian pharmaceutical companies over the past three years:

  • Ranbaxy, in which Daiichi Sankyo acquired 64% stake in June 2008, with an implied enterprise valuation of $8.5 billion (27 times EBITDA multiple), was hit by a broad import ban by the FDA of 30 generics products from two production sites at Paonta Sahib and Dewas in September 2008, issues which are yet to be resolved. This has led to the resignation of two consecutive Ranbaxy chief executive officers, and its chief financial officer/President, as well as reassignments within Daiichi. Resolving these issues created real worries for Ranbaxy and Daiichi Sankyo, since Ranbaxy held 180-day exclusivity on the U.S. generic sales of Lipitor starting Nov. 30, 2011. This six-month period could have added as much as $600 million in sales to Ranbaxy, but was jeopardized by the import ban. Ranbaxy’s shares trades 30% below Daiichi’s purchase price three years ago. The company ultimately received last-minute clearance from the FDA, but Ranbaxy cut a 50/50 profit-sharing partnership with Teva as part of the process.
  • Sanofi bought an 80% stake in Shantha for $600 million in 2009, only to see its pentavalent vaccine Shan5 broadly recalled by the WHO in 2010, wiping out more than $340 million in sales for that year.
  • Pfizer had to recall IV bags produced by Claris Lifesciences with which it has a deal for the supply of generic injectables, and Claris was subsequently hit by an import ban by the FDA in 2010.
  • A similar situation arose with Aurobindo, which was hit with an import ban in 2011 due to problems at its antibiotics plant.

Nonetheless, Divi’s Laboratories, Dr. Reddy’s, Glenmark, Hikal, Lupin, Sun Pharma and Zydus Cadila are leaders in generics supply to the U.S. market and India holds 1/3 of approved ANDAs in any given year by the FDA.
Another example concerning quality, which is particularly pressing in the life science industry, is the steroid case mentioned in the sidebar at the end of this article. Steroids are complex molecules made by a combination of sophisticated traditional and biotech processes. They have very demanding specifications and analytical method requirements. A further aspect is the “end of pipe analytic” in the production chain. It is a challenge to develop an analytical method for detecting all by-products in these complex molecules. The whole production chain has to be controlled carefully to avoid risks. Furthermore, the dosage is very low so that the incidence of the product cost is almost negligible. This, plus the pressure from stakeholders of GlaxoSmithKline and Pfizer to keep the plants running — and not simply cost considerations — are the basic reasons for insourcing steroids production back from China.
For the sake of good order, it must be stated that even western big pharma is not immune to quality issues. As a matter of fact, eight out of the top 10 pharma companies, all headquartered in the U.S. or EU, received warning letters from the FDA in 2010. With 12 letters, Johnson & Johnson led the list.
As part of Big Pharma’s ongoing restructuring initiatives, outsourcing of API manufacturing has been gaining ground compared with in-house production. Primarily because of their labor cost advantage, Chindian fine chemical companies have been successful at capturing a rapidly growing part of the demand, which was also fostered by the patent cliff. This situation will essentially persist in the future:
The difference in labor costs will gradually narrow down. The complex interaction of labor wage increases with higher productivity in Chindia will be such that we expect to reach some form of labor parity in the coming 20 years (see Figure 1). CapEx requirements for new plants will remain to the advantage of Chindia.
Chindia is creating a higher number of scientists compared to the West: for instance, India trains annually six times the number of chemists trained in the U.S. Taking Intellectual Property as a benchmark for the innovation, Chindia’s 350,000 patent applications in 2010 are second only to the U.S.A. (450,000 [2010])8. (Of course, there are questions of how well trained these scientists are and how pertinent these patents will prove.)
All in all, claims by Western managers over the fading competitiveness of Asian players vs. Western counterparts are more fiction than fact.
With regard to the soft issues, total cost of ownership has to be taken into consideration9. In contrast to the pure costs of products and services, it is likely to stay higher in the case of complex technology transfers or small size deals.
According to GlaxoSmithKline’s Sir Andrew Witty, the high labor cost countries (HLCC) can compete with low labor cost countries (LLCC), given three basic premises apply to the production site:10

  • Lean site with minimal overhead
  • Relentless focus on continuous operational and process improvements
  • (Nearly) fully loaded production site.


The future development of China’s and India’s competitiveness is ambiguous: In terms of technologies, non-conventional manufacturing processes will gain importance. Within small molecules, the forthcoming new drugs will increasingly be low volume/chiral/high potency/high toxicity. Flow reactors and high containment equipment will partially replace the conventional batch reactors and sophisticated chromatographic methods the conventional purification by fractional crystallization. As these processes are automated and the equipment — micro-reactors, glove boxes, simulated moving bed columns — must be sourced often from HLCC, the competitive advantage of LLCC will diminish. It has to be kept in mind, however, that the share of fine chemicals made by these technologies, although growing, will stay small. Possession of a new technology is not in itself a route to profit11.
For large molecules, the key success factors are the mastering of the operational challenges of the mammalian cell technology used for the synthesis of biopharmaceuticals and biosimilars, primarily high containment, reduction of the comparably low time/volume output and the demanding analytical methods. The winners will be the innovative companies discovering quantum leap improvements in the standard operation costs, independently of the geographic location of the plant site.
In business terms, there are no doubts that Asian producers outperform western counterparts based on production cost, except in cases where there is a large disparity between total cost of ownership and product cost, such as for small orders of custom-made products. Furthermore, it has to be considered that TCO targets are usually not included in the incentive programs of procurement managers in pharma companies. They typically create incentives based on savings generated from the API purchasing price. This one-dimensional metric slants the equation towards Chindia. It does not catch management’s attention, at least as long as no major problem arises.
Last but not least, fine chemical companies in LLCCs will be the main beneficiaries of the booming local demand for Western medicine. Actually, the major demand-pull for both drug substances and drug products will come from developing countries. It is hardly conceivable that producers from HLCCs will be able to participate in this booming market from a western-based production platform.



  1. H.L. Sirkin, M. Zinser and D. Hohner, Made in America, Again, Boston Consulting Group,  Boston (2011), 15 pp.
  2. M. McCoy, Taking back production from Asia, Chemical & Engineering News, Apr. 27, 2009,  pp.16-17
  3. Stefan Borgas, Der Standort Visp ist profitabel, hoch kompetent und vital, Chemie Plus, Sep. 2011, pp. 4-10
  4. B. Kennedy & M. Baumann (Thomson Reuters), API Sourcing in China and India – is Asia the Only Option for the Future?, CHEManager, Oct- 2011, p. 15
  5. Peter Pollak, Will raising labor costs impact the competitiveness of the Indian Fine Chemical Industry?, Chemical Weekly, Feb. 24, 2009, p. 183
  6. Peter Pollak, Fine Chemicals – The Industry and the Business, 2nd edition, John Wiley & Sons, Inc. Hoboken NJ (2011), 280 pages, ISBN 978-0-470-62767
  7. The Division of Science Resources (SRS) of the National Science Foundation; U.S. Institute of Applied Manpower Research, India (2011)
  8. World Intellectual Property Report 2011, WIPO, Geneva, ch. 1, p. 53
  9. Daniela Hoffmann, Eine Kosten-/Nutzenanalyse des Einkaufs chemischer Rohstoffe aus China für die pharmazeutis-che Wirkstoffproduktion, Bachelor-Arbeit, Fachhochschule Mainz. Feb 08, 2010
  10. Andrew Badrot, High Labor Cost Countries are making a comeback against low cost counterparts, Contract Pharma, Sep. 16, 2011
  11. Ian Grayson, The Madness of Fine Chemicals, Chimica Oggi – chemistry today, Jan/Feb 2012, (in print)


Pro-insourcing statements of western life science companies


Statements of Western life science industry executives regarding the impact of escalating labor costs on the competitiveness of the Asian fine chemical industry
Andrew Witty, CEO of GlaxoSmithKline, announced in April 2011 that his company would bring steroid bioprocessing back from India to GSK’s Montrose, Scotland facility, a site scheduled for shutdown. In July 2011 he said, “The west needs to realize that the future has to include manufacturing, and we ought to be challenging ourselves to demonstrate that western sites can be absolutely the most efficient and innovative”
In a similar move, Pfizer CentreSource, the manufacturing arm of the drug giant had rethought its plan to outsource part of the steroid production at its Kalamazoo, USA plant to two Asian firms, China’s Shanghai Pharmaceutical Group and Taiwan’s ScinoPharm to CM Scinopharm in Taiwan to its Kalamazoo plant.2
In a recent interview with Chemie Plus Stefan Borgas, CEO of Lonza, said, “All in all, Asian players do not produce cheaper than Western API producers. Admittedly, e.g. Indian producers pay one third of Lonza’s cost for the construction of one cubic meter of reactor volume, but the output is much lower. Per cubic meter reactor volume, we can convert up to 10 times more than our Indian competitors in their plants.”3 (translation from German)
Luca Mantovani, President DSM Pharma Chemicals: “There is still a large gap in hardware and software comparing with western players although Asia is catching up to meet the standards of FDA and EMA  but this all process will drive costs up.”
Note: For a discussion of these comments, see “Total Cost of Ownership”


Pro-offshoring statements of western life science companies


Novartis closes manufacturing site for its top selling drug, Diovan, in Basel (Switzerland) and transfers chemical production to a LLCC.

Neue Zürcher Zeitung, Oct. 31, 2011, p. 9
AstraZeneca invests $200 million in a new manufacturing facility in China Medical City, Taizhou.

– AZ press release Oct. 10, 2011
Lonza closes the Wokingham (UK) and Shawinigan (CA) sites and divests the plant at Conshohocken (Riverside), PA.

– Lonza Activity Report 2010, p. 4
Evonik started-up a facility to manufacture APIs in Nanning, province of Guanxi

– Annual Report 2010, p. 51
Lonza is investing $200 million in Nansha for a new multipurpose API and intermediates plant complex.

– Lonza Activity Report 2009, p. 4



The authors would like to thank Gian-Paolo Negrisoli, president of Flamma SpA, Anish Swadi, senior vice president, business development of Hikal Ltd., and Guy Villax, chief executive officer of Hovione FarmaCiencia SA for their input and guidance.


Andrew Badrot is the founder and chief executive officer of CMS Pharma. He can be reached at info@cms-pharma.comPeter Pollak, Ph.D. is an independent consultant and a board member of various fine chemical companies. He can be reached at peterpollak@swissonline.chDr. Rolf Dach is an independent consultant to fine chemical companies. He can be reached at rotego-chem@t-online.de.

Leave a Reply

Your email address will not be published.