Human capital in personal consumption and labour force productivity: evidence from an OECD panel

Abstract

The relationship of household stocks of information cumulated from personal consumption to a labour productivity index (LPI) is investigated. The regression model in an OECD panel controls for effects of domestic and foreign R&D, educational level and institutional factors on LPI. Human capital is proxied by mean years of education, and stocks of household information are constructed from deflated expenditure on a set of information goods in personal consumption. After unit root and cointegration testing, models were estimated with dynamic ordinary least squares (DOLS). Results show that the stocks of information constructed from consumer expenditure flows are related to LPI after control variables are accounted for. Effects of LPI on the stock of information are also investigated. In these results, the stock of information had a larger effect on LPI than LPI had on the stock. Policy implications for productivity objectives that include differential tax rates on consumption of information and conventional goods are noted.

Keywords:

• human capital
• consumption
• growth
• household production

JEL Classification:

• E21
• J24
• O47

Notes

¹ Information accumulated in home activities can range from a standard of abstract generalizations that some require as definitional, to more cursory ‘facts’ that households retain from their activities. Although the standards for knowledge remain somewhat controversial, there is a basis to expect that basic properties of information as nonrival and slowly depreciating across a wide range of definitions are sufficient to develop the fundamental arguments even at what may be a lower bound of its abstractness and generality. In fact, it is surprising how cursory information can be and still show effects of its use (e.g. Purcell-Gates, 1996).

² A major limitation of the many unit root tests is the assumption that all groups in the panels have the same parameter. The Im-Pesaran-Shin (IPS: 2003) test relaxes the assumption of a common parameter and instead allows each panel to have its own parameter. The null hypothesis for IPS is that all panels have a unit root (H0: rH0_i = 0 for all i). The alternative hypothesis is that the fraction of panels that are stationary is nonzero. Fisher-type tests for panel-data unit roots conduct unit-root tests for each panel individually and then combine the p-values from these tests to produce an overall test. While the above tests have as the null hypothesis that the data contain a unit root, the Hadri test statistic has the null and alternative hypotheses are reversed. Significant test statistic indicates the presence of a unit root and a variable being nonstationary I(1).

Additional information

Funding

The author gratefully recognizes Funding support from the Sally and Donald Lucas Foundation.