Cultural intelligence for leaders: the four CQ capabilities

You are about to sign a first order with a foreign distributor, and next quarter your team gains three people of three nationalities. Should you read up on your partner's country, or build something else? Cultural intelligence (CQ) answers the second half: what you build is a set of your own capabilities, not a table of traits for a people.

What are the four capabilities of cultural intelligence?

Ang et al. define CQ as an individual's capability to function and manage effectively in culturally diverse settings. Earley and Ang developed the concept in their 2003 book (Ang et al., 2007; Earley and Ang, 2003). The four parts follow Sternberg's framework: three capabilities sit inside the head, and one is outward action (Ang et al., 2007).

Metacognitive CQ is planning, monitoring and revising your mental model of a group's norms, including questioning your own assumptions. Cognitive CQ is knowledge of customs, economic and legal systems, and frameworks of cultural values such as Hofstede's. Motivational CQ is directing attention and energy towards learning and working across differences, driven by real interest and confidence that you can do it. Behavioural CQ is using suitable words and non-verbal actions, from tone to gesture (Ang et al., 2007).

Which capability should you build first?

The table is my own self-check. It is not the 20-item CQS scale and gives no score. Using the official scale outside academic research needs permission from its rights holder, whose contact the paper gives (Ang et al., 2007).

Any row where you answer "not yet" is a candidate. If there are several, I would start with metacognitive CQ. That is my reading, not the conclusion of Rockstuhl and Van Dyne. My basis: in samples with a high average metacognitive CQ, behavioural CQ was far more strongly related to adjustment (coefficient 0.30, against 0.01 in low samples) (Rockstuhl and Van Dyne, 2018). The same paper found the shared core of CQ was less strongly related to adjustment in high samples (0.20, against 0.61). These compare samples, not individuals, and the authors say results may differ at the individual level, so treat them as a hint. The table suits you if you only have time to build one thing. It does not suit you if you need a score to select candidates.

A packaging order with a foreign distributor: an invented example

This example is invented: the firm, the people and the figures are mine. I do not name the partner's nationality, so as not to invite a stereotype.

Long Chau Packaging (invented) is a carton factory in Can Tho. Ms Linh, the sales director, asks a new distributor to cut payment terms from 60 days to 45. Mr K., the distributor's purchasing head, replies: "We will look into it." She writes in the minutes that the partner agreed, and builds her cash plan on 45 days. Two weeks later the first order, 400 million dong, is confirmed on 60-day terms. If orders stay at 400 million dong a month, 15 days is 200 million dong of extra working capital tied up, counting a 30-day month: 800 million at 60 days against 600 million at 45.

"We will look into it" could mean yes, a polite no, or waiting for approval from above. Ms Linh picked the reading that suited her, and the fault is not a lack of knowledge about Mr K.'s country. Ang et al. argue that people who notice possible differences in how others think tend to read behaviour from the actor's point of view (Ang et al., 2007). The four moves below are my picks to illustrate each capability, not research results:

• Metacognitive: write down the assumption "look into it means yes" next to two other readings, and treat the terms as unsettled.
• Cognitive: ask who at the partner can approve payment terms, and how long approval takes.
• Motivational: be ready to add another call rather than treat the round as finished.
• Behavioural: send a summary email with a question that can be answered yes or no, and ask "what would you need to see to settle on 45 days?"

How does CQ differ from national culture dimensions?

CQ is a capability of one person. GLOBE culture dimensions describe the culture of a society. They sit at different levels, and in my view mixing the levels is how stereotypes start.

GLOBE was a ten-year research programme. A team of 160 scholars, working from 1994, surveyed over 17,000 middle managers in three industries across 62 cultures, using dimensions such as power distance and uncertainty avoidance (House et al., 2004). Ang et al. place knowledge of cultural value frameworks inside cognitive CQ (Ang et al., 2007). In my reading, GLOBE results are raw material for cognitive CQ and cannot stand in for all four capabilities.

Li et al. asked whether national culture goes with differences in the link between leadership style and employee engagement. They pooled 209 studies, 82,386 people and 45 countries, and assigned GLOBE scores by the country where each survey took place, since the original studies rarely measured culture. Of 56 tests, 10 were statistically significant, and some ran against the authors' own predictions: transformational leadership was more strongly linked to engagement in countries with low gender-egalitarianism scores (0.50) than in those with high scores (0.32), meaning one standard deviation below and above the average. The paper measures employee engagement, not negotiation, and 79 per cent of its studies were cross-sectional, so it cannot establish cause (Li et al., 2021). A related idea is country differences in leadership effects.

What does CQ predict, and what has the evidence not said?

Ang et al. tested the model in three studies of 794 people in the USA and Singapore: metacognitive and cognitive CQ predicted intercultural judgement; motivational and behavioural CQ predicted adjustment; metacognitive and behavioural CQ predicted task performance (Ang et al., 2007). Rockstuhl and Van Dyne rechecked this on 199 samples and 44,155 people, adding outcomes rated by other people to avoid bias from a single source. They found:

• Judgement: metacognitive (coefficient 0.18) and cognitive (0.12) CQ predict it; motivational and behavioural are near zero.
• Adjustment: motivational CQ predicts it (0.28 for sociocultural adjustment, 0.26 for psychological well-being). Behavioural CQ was not significant, contrary to Ang et al.'s expectation.
• Task performance rated by others: behavioural (0.28) and metacognitive (0.11) are positive. Cognitive is negative (-0.10) once the shared core and the other three are controlled, although its simple correlation is positive; the authors call this unexpected (Rockstuhl and Van Dyne, 2018).

The shared core of the four capabilities has a positive coefficient for all seven outcomes. For adaptive performance, the four specific capabilities add nothing beyond it (Rockstuhl and Van Dyne, 2018).

What has the evidence not said? Both papers report statistical associations, and I saw no experiment showing that training CQ improves results. Ang et al. sampled in the USA and Singapore and say future work should cover Europe, Africa, Latin America and the Middle East (Ang et al., 2007). In the parts I read, neither paper reports results for Vietnamese people or firms. The CQS is a self-assessment, so a score reflects how a person sees themselves; that is my inference from the instructions given to respondents (Ang et al., 2007). Results vary considerably between studies, so the boundary conditions of CQ are unclear (Rockstuhl and Van Dyne, 2018).

Going deeper: where does CQ come from, and what are the criticisms?

Earley and Ang published the book in 2003 (Earley and Ang, 2003). I read only a bookseller's description of it, not the book. That description says the book develops CQ through three facets: cognition (recognising patterns from cultural cues), motivation and behaviour (Earley and Ang, 2003). Ang et al. attribute the four-dimension model, including metacognition, to Earley and Ang (Ang et al., 2007). Rockstuhl and Van Dyne credit Ang et al. (2007) with delineating the four-factor model, and cite Earley and Ang (2003) when describing its Sternberg basis (Rockstuhl and Van Dyne, 2018). I cannot say where metacognition sits in the book; this article follows Ang et al.'s account. The CQS has 20 items and was validated in Singapore and the USA (Ang et al., 2007).

• One factor or four? Rockstuhl and Van Dyne propose a bi-factor model: one general CQ factor plus four specific capabilities. It fits the data better than a one-factor model for six of the seven outcomes (Rockstuhl and Van Dyne, 2018).
• Not fully replicated. The authors say research before 2018 had not replicated Ang et al.'s initial evidence that the four capabilities predict different outcomes; their synthesis supports it in part and departs on behavioural CQ and adjustment (Rockstuhl and Van Dyne, 2018).