---
title: "Dunning's OLI Paradigm: Invest Abroad or Export?"
description: "OLI needs all three advantages (ownership, internalisation, location) before you invest abroad, not \"one or more\". With Tesla, Green SM and Starbucks."
url: "https://staging.khoi.co.uk/en/blog/oli-dunning-du-ca-ba-cua"
lang: en-GB
author: "Lê Quốc Khôi"
date: 2026-11-03
updated: 2026-11-03
topics: ["Strategy & leadership"]
tags: ["OLI paradigm", "Foreign direct investment", "Entry modes", "Internationalisation"]
---

# Dunning's OLI: invest abroad or export?

## Key takeaways

- Dunning (1988) ties the extent, form and pattern of international production to the configuration of three sets of advantages: ownership, internalisation and location (Dunning, 1988, p. 2).
- Saying "one or more" advantages is enough is wrong for new investment; the site's library records that the standard statement requires all three [Dunning's eclectic (OLI) paradigm](https://staging.khoi.co.uk/en/framework/oli-paradigm). Dunning writes that withdrawal needs only one of the three OLI variables to be absent (Dunning, 1988, p. 22).
- The shortcut "no location advantage, export; no internalisation advantage, license" is the library's, not a result Dunning proves: he says little headway has been made in unifying explanations of trade and production (Dunning, 1988, p. 18).
- In 2019 Tesla was, according to Xinhua, the first to use a policy letting foreign carmakers set up wholly owned subsidiaries in China: a government can open or close a gate (Xinhua, 2019).
- Starbucks closed 61 of its 85 shops in Australia but, according to an expert from the University of New South Wales quoted by the BBC, did very well in Japan and China (BBC News, 2008).

On 30 July 2026 Green SM launched an electric taxi service in Copenhagen. The company's own release says it owns and manages the fleet instead of only connecting passengers with independent drivers (Green SM, 2026). You cannot put a taxi in a container bound for Denmark, so Green SM never faced a choice between investing and exporting. A furniture maker does, and a board will often settle it with one sentence: "We have an edge, so we invest."

This post asks whether one edge is enough. The [OLI paradigm](https://staging.khoi.co.uk/en/framework/oli-paradigm) says it is not. How exporting, licensing and franchising compare is covered in the post on [choosing a foreign market entry mode](https://staging.khoi.co.uk/en/blog/phuong-thuc-tham-nhap-thi-truong); here I go deeper on reading the conditions for investing.

## What do Dunning's three advantages ask?

Three advantages make three separate questions.

Gate O, ownership, asks what you hold that rivals in the target country lack. Dunning (1988) says the advantage must be enough to cover the cost of setting up and running a foreign operation, on top of costs that local producers do not carry (Dunning, 1988, p. 2).

Gate I, internalisation, asks whether doing it yourself beats selling the right. This is the second condition: the firm must see more benefit in moving its advantage across a border inside its own organisation than in selling it, or the right to use it, to foreign firms (Dunning, 1988, p. 3).

Gate L, location, asks why produce there at all. A firm goes abroad when it gains by combining intermediate products that can travel from home with at least some immobile resources, or other intermediate products, in the other country. Trade barriers have historically drawn in a good deal of manufacturing investment, and government intervention can encourage or discourage inward investment (Dunning, 1988, p. 4).

> **Figure:** [Dunning's eclectic (OLI) paradigm](https://staging.khoi.co.uk/en/framework/oli-paradigm) (redrawn from the site's library framework)

## One advantage, or all three?

All three, if you mean to invest in production abroad. One advantage only earns you the right to ask the next question.

The 1988 paper has no sentence saying outright that all three are needed for a first entry, but it supports the strict reading. Ownership is the first condition for competing by producing in another country (Dunning, 1988, p. 2), internalisation is "the second condition" (Dunning, 1988, p. 3) and location is the third strand (Dunning, 1988, p. 4). The closest evidence is the passage on divestment: withdrawing differs from first entry because it requires the absence of only one of the three OLI variables (Dunning, 1988, p. 22).

My suspicion is that "one or more" is the withdrawal rule turned round to face entry. I could not trace who first put it that way, so that is a guess. The consequence is not a guess. If one advantage were enough, every firm with a strong home brand would have a reason to build a plant. If all three are needed, a strong brand with no reason to produce locally still points to exporting.

## What if a gate is missing, and how sure can you be?

The library's flow diagram reads: no O, no basis for investing abroad; O but no L, export from home; O and L but no I, license or partner [Dunning's eclectic (OLI) paradigm](https://staging.khoi.co.uk/en/framework/oli-paradigm). That works as a first screen. It does not work as a theorem, for two reasons found in the 1988 paper itself.

First, Dunning admits some progress with joint ventures and non-equity contracts, but only limited headway in unifying explanations of trade and production (Dunning, 1988, p. 18). The "no L, so export" branch touches exactly that gap, and "license or partner" is the library's compact reading, not something Dunning proves.

Second, "advantage" changes meaning with the motive for investing. The parameters drawing a firm into a copper mine in New Guinea are unlikely to match those of a Japanese colour television maker investing in the United States (Dunning, 1988, p. 5). State the purpose before you score any gate; the paper's Figure 2 separates market-seeking, resource-seeking and efficiency-seeking production (Dunning, 1988, p. 13).

## Can a government open or close a gate?

Yes, and there are two documented examples. Dunning describes Japan: in OLI terms, it at first did not let foreign multinationals internalise most markets in intermediate products, technology above all; it acquired them in other ways or promoted home-grown production (Dunning, 1988, p. 17). Gate I was locked by policy against outsiders.

The second goes the other way. On 7 January 2019 Tesla broke ground on its Shanghai factory. Xinhua called Tesla the first to benefit from a new policy allowing foreign carmakers to set up wholly owned subsidiaries in China, while BMW had announced in October 2018 that it would raise its stake in its Shenyang joint venture from 50 to 75 per cent (Xinhua, 2019). The details are in [the Tesla Shanghai case](https://staging.khoi.co.uk/en/case/tesla-shanghai-gigafactory).

On gate L the report gives one datum: the Model 3 started at USD 35,000 in the US and sold for roughly 500,000 to 560,000 yuan in China, and local production was expected to bring the price down (Xinhua, 2019). It does not explain the gap, so I do not say how much is duty or freight. On gate O it mentions Tesla's charging network and Beijing R&D centre in China but does not call them an ownership advantage, so I leave that cell blank.

## What does OLI ask of a service that cannot be exported?

Gate I again: run it yourself, or let an independent partner do it. Green SM oversees driver operations to keep service quality consistent (Green SM, 2026), which fits one motive for internalising that Dunning lists, ensuring the quality of the end product (Dunning, 1988, p. 3). The paper also records the opposite pull: the administrative costs of hierarchies, and external diseconomies of operating abroad (Dunning cites the Bhopal disaster), make contracts, or at least a jointly shared equity stake, more probable (Dunning, 1988, p. 3).

The release does not say whether the company weighed franchising, and it does not name an ownership advantage (it mentions only experience of serving millions of journeys across Asia). It is the company's own document, so I read it as what the company wants announced, not as evidence of its internal reasoning. See also [the Green SM case](https://staging.khoi.co.uk/en/case/green-sm-market-entry).

## What does leaving need? Starbucks in Australia

Dunning predicts that firms will want to reduce their presence in a country in two situations: their competitive advantage over local firms weakens, or the balance between using outside markets and internal hierarchies shifts (Dunning, 1988, p. 22).

Eight years after it began selling coffee in Australia, Starbucks had closed 61 of its 85 shops, the BBC reported on 4 August 2008. A consumer psychologist in Melbourne said the chain was perhaps too standardised, was quickly copied and lost its lustre (BBC News, 2008). Through OLI that is a gate O question: does the advantage still cover its costs when local rivals do something close? That is my reading of an expert's opinion, not a measurement. Analysts also said the company expanded too fast and took on too much debt, which sits outside OLI (BBC News, 2008).

The detail worth keeping: John Roberts of the University of New South Wales told the BBC that Starbucks had done very well in international markets with no traditional coffee-drinking culture, and he pointed to Japan and China (BBC News, 2008). One company, doing well in one country and closing shops in another: OLI is scored country by country, not company by company. The full case is [Starbucks leaves Australia](https://staging.khoi.co.uk/en/case/starbucks-australia); another exit, by acquisition, is in the [Walmart in Germany post](https://staging.khoi.co.uk/en/blog/walmart-o-duc-thich-nghi-sai).

## How should a board use it?

I built the table below from the conditions Dunning sets out, adding a policy row drawn from the Japan and China examples. The two right-hand columns are my suggestions, not Dunning's.

| Gate | Question for the board | Evidence to have on the table | Reason to stop |
|:--|:--|:--|:--|
| O | What do we hold that local firms lack and cannot buy? | A process or system already running in at least one other country | The edge exists only at home: connections, wage levels |
| L | Why produce on the spot rather than at home and ship? | The share of price that is border cost, resources found only there | The sole reason is "a big market" |
| I | If a local partner does it, what do we lose? | Risk of the process leaking, how far contract terms can protect it | Nobody has priced a good licence |
| Policy | Which forms does the target country allow? | Open sectors, maximum ownership share, expiry dates | The rule has just changed or is about to |

This suits a board that is discussing building or buying a site and has heard someone say "we already have the edge". It does not suit a first trial shipment, where the loss you can bear is the real question.

## Going deeper: what did I read, and what not?

I read Dunning (1988) in full from the journal's scanned copy. The 1980 paper, also listed by the library as a source, I have not read. In the 1988 paper, no table pairs each OLI combination with exporting, licensing or investing: that pairing comes from the library. The library also lists four investment motives, including strategic asset seeking; the 1988 paper names only three in Figure 2 (Dunning, 1988, p. 13), so I do not use the fourth.

Each case rests on one public source that I read in full. Xinhua's text sits on the Pudong district government's site, so it is Chinese state media; Green SM's release was written by the company; the BBC piece is mostly analysts' opinion. None gives profit or loss by country, so this post does not say which decision was right.

## Frequently asked questions

### What does OLI stand for, and why is it called eclectic?

Dunning (1988) says he first put forward the eclectic paradigm of international production in 1976, at a Nobel Symposium in Stockholm. "Eclectic" was meant to convey that a full explanation of firms' transnational activity has to draw on several strands of economic theory, and that foreign direct investment is only one of several channels of international economic involvement (Dunning, 1988, p. 1). O, L and I are the initials of the three sets of advantages: ownership, location and internalisation (Dunning, 1988, p. 5).

### Should the three gates be taken strictly in the order O, L, I?

The order in the diagram is a convenience for asking questions. Dunning writes that the decision on where to site a factory is not independent of who owns the assets, nor of the route by which the assets or the right to use them are transferred (Dunning, 1988, p. 4). If your answer at one gate changes, go back and rescore the others.

### Can OLI predict what a particular company will choose?

No, and Dunning says so plainly: being general, the paradigm has only limited power to explain or predict particular kinds of international production, and even less the behaviour of individual firms (Dunning, 1988, p. 1). He cites Horst (1972) when noting how little empirical work there is on firm-level behavioural variables: apart from size, Horst could identify no firm-specific variable that satisfactorily explained the degree of multinationality of US-owned firms (Dunning, 1988, pp. 7-8). Use OLI to put questions to your own decision, not to guess at rivals.

### Does an SME with its first foreign order need to score OLI?

Usually not yet. For a trial shipment the questions are how much you can lose without pain and which small step comes next; OLI matters once someone proposes building or buying a site abroad. That is my judgement, and I have no data on Vietnamese SMEs.

## References

- BBC News (2008) *Shunned Starbucks in Aussie exit (4 August 2008)*. BBC News. Available at: http://news.bbc.co.uk/2/hi/7540480.stm.
- Dunning, J.H. (1988) ‘The eclectic paradigm of international production: a restatement and some possible extensions’, *Journal of International Business Studies*, 19(1), pp. 1–31. https://doi.org/10.1057/palgrave.jibs.8490372
- Green SM (2026) *Green SM officially launches fully electric taxi service in Denmark*. Green SM: news. Available at: https://www.greensm.com/dk-en/news/green-sm-officially-launches-fully-electric-taxi-service-in-denmark.
- Xinhua (2019) *Report on the groundbreaking of Tesla's Shanghai factory (7 January 2019)*. Shanghai Pudong government English site. Available at: https://english.pudong.gov.cn/2019-01/08/c_314114.htm.

---

Canonical HTML version: https://staging.khoi.co.uk/en/blog/oli-dunning-du-ca-ba-cua