The share market boom in the 1990s is often linked to the acceleration in labour and total factor productivities over the same period. This paper explores the argument that labour and total factor productivities are inaccurate measures of firm's earnings, which underlie equity valuations, and that capital productivity is a better measure of earnings. Using 80 years of data for 11 OECD countries, it is shown empirically that the link of capital productivity to share returns is indeed stronger than that of labour productivity and TFP.
The worldwide increase in equity prices in the 1990s has often been linked to accelerating labour productivity, which has in turn been related to the information technology (IT) revolution (see the discussions in Campbell and Shiller, 2001, Greenwood and Jovanovic, 1999, Hobijn and Jovanovic, 2001, Keon, 1998, Hall, 2001, International Monetary Fund (IMF), 2000 and Madsen and Davis, 2006). It has been argued that the apparent acceleration in labour productivity is an accurate indicator of increases in firms' current and expected real earnings and thus dividends, and its acceleration has thereby contributed to an increase in the value of firms.
The contribution of this paper is twofold. First, it is shown that labour productivity and to a lesser extent total factor productivity (TFP) as well as potential output are misleading proxies for firms' earnings. It is shown that the more relevant productivity measure for the remuneration of capital, and hence a key element for proxying firms' earnings, is the marginal productivity of capital, which under certain assumptions is equivalent to capital productivity at a macro-level. Second, the relationships between share returns and various productivity measures are empirically investigated using historical data on share returns and productivity over 80 years for 11 countries (G7, Australia, Netherlands, Sweden and Denmark). Short run and long run Granger causality tests are first used to examine the bivariate relationship between real share returns and various productivity growth measures. Thereafter, the nexus between real share returns and productivity is investigated using a vector-error-correction-mechanism (VECM), where other variables than productivity are allowed to influence share returns. Finally, panel data estimates using similar variables are presented to exploit the cross-country correlation between the error terms. All results underpin the case for using capital productivity as a measure for firms' earnings and hence as a predictor of equity returns.
The paper is organised as follows. Section 2 establishes a model of firms and consumers to show that share prices are linked to the marginal productivity of capital, the bond yield and the equity risk premium. Section 3 shows why other macro-indicators of productivity, such as growth in labour productivity and to some extent also TFP, are misleading indicators of earnings. Empirical estimates are presented in Sections 4 and 5. They show that capital productivity is indeed the measure which is most closely linked to share returns.
The share market boom in the 1990s is often linked to the accelerating labour productivity over the same period. In this paper we have shown that labour productivity and TFP are misleading measures of corporate earnings. The problem associated with growth in labour productivity as a measure of earnings is that it is heavily influenced by capital deepening which lowers earnings per unit of capital due to diminishing returns to capital. TFP growth influences earnings positively, but because TFP growth is often triggered by capital deepening which embodies new technologies (Wolff, 1991), the nexus between stock returns and TFP growth is distorted. Based on a general equilibrium model it was shown that capital productivity is the correct measure of earnings, because capital is paid its marginal product in equilibrium.
The nexus between stock returns and various productivity measures was examined using 80 years of data for 11 industrialised countries. Due to changing dividend payout ratios, we use the cumulated total stock return as our measure for equities and not the share price index. Of the three measures of productivity, total stock returns were most strongly related to capital productivity. Total stock returns tend to lead productivity in the short term, consistent with the forward-looking nature of equity markets. VECM variance decompositions show strong linkages from capital productivity to total stock returns in the short and long run in virtually all countries, which exceed those from labour productivity and TFP. Meanwhile, impulse responses suggest that capital productivity tends to have consistent positive effect on total stock returns at least up to 10 years, which is less present for labour productivity or TFP. These results were reinforced by panel data estimates, suggesting that the findings were very robust to estimation method and model specification.
In opposing the view that labour productivity is an appropriate measure of equity market fundamentals, our results are consistent with the findings of Campbell and Shiller (2001), who sought to assess the argument that the high stock market value is often justified by expectations of a continuation of the high labour productivity growth in the 1990s, with an underlying premise that labour productivity is the relevant productivity measure of earnings. In fact, finding that price-smoothed-earnings ratio cannot predict future labour productivity for the US, they concluded that the high share prices at the time of writing could not be due to a rational forecast of labour productivity growth.13
