• SOCIAL CAPITAL
• SELF-RATED HEALTH
• MENTAL HEALTH

I sincerely appreciate Dr Ramiro Manzano Nunez's comments on my article,1 which examined how the association between individual-level social capital and health is confounded by an individual's unobserved time-invariant attributes. He stressed the importance of linking social capital in addition to bonding and bridging social capital, which I addressed in my article. He suspected that the lack of this component could confound my results.

I agree that the linking aspect of social capital is an important concept. Linking social capital represents “norms of respect and networks of trusting relationships between people who are interacting across explicit, formal or institutionalised power or authority gradients in society.”2 It is different from bonding and bridging in that it is concerned with relations between people who are not on an equal footing.

My article did not treat linking social capital, because “The Longitudinal Survey of Middle-Aged and Older Adults,” on which my statistical analysis was based, did not address it. This was not on account of my neglecting its potential importance.

However, the point is whether my results would be affected by including linking social capital in the analysis. I cannot directly answer this question, because I have no data on linking social capital. However, I suspect that the results would remain virtually the same.

The main finding of my study is that the association between social capital and health tends to be overestimated unless an individual's unobserved time-invariant attributes are controlled. This is the most likely outcome because social capital is an endogenous variable for an individual, especially if it is assessed at the individual level. To access and/or make use of social capital—whether bonding, bridging or linking—is not fully an external shock to an individual; it usually involves an individual's action—whether active or passive—and hence is likely affected by his/her attributes, not only time-variant but also time-invariant ones.

My results supported this view for bonding and bridging social capital. I can reasonably predict that the same is true of linking social capital, which is endogenous like the other two types of social capital and hence likely affected by an individual's attributes. Nonetheless, I agree that it is of great interest to compare the results across all three types of social capital.