Academic journal article Contemporary Economic Policy

Estimating Productivity Growth in the Korean Economy without Restrictive Assumptions

Academic journal article Contemporary Economic Policy

Estimating Productivity Growth in the Korean Economy without Restrictive Assumptions

Article excerpt

This study eliminates the effects of markup, returns to scale, and capital utilization from the Solovv residual (SR) for Korea to derive an alternative measure for productivity. Empirical results show that markup generates significant bias in the SR, and that the alternative productivity measure is greater than the residual. Furthermore, money supply Granger-causes the SR but does not Granger-cause the alternative measure, suggesting that the new productivity measure is consistent with the neutrality of money. The results contradict the presumption that the SR with variable capital utilization represents true technology shocks because it is orthogonal to demand shocks. (JEL C32, E32, 047)

I. INTRODUCTION To estimate the Solow residual (SR), researchers generally assume constant returns to scale (CRS), perfect competition, and the full employment of factor inputs. The SR fails to provide an accurate measure of total factor productivity (TFP) change if one of these assumptions is violated. For example, previous studies have found that market power and increasing returns to scale (RTS) generate procyclical noise in the SR (Caballero and Lyons 1992; Hall 1988, 1989). Other studies have reported that cyclical factor utilization is the most likely cause of the mismeasurement of technology shocks, as measured by the SR (Basu 1996; Burnside and Eichenbaum 1994; Mankiw 1989; Paquet and Robidoux 2001; Sbordone 1996; Summers 1986).

One prominent feature of the SR as measured under the restrictive assumptions is that the residual varies with demand conditions, tainted by procyclical noise. However, many studies have eliminated this cyclical bias using flexible capital utilization (e.g., Basu 1996; Burnside, Eichenbaum, and Rebelo 1995; Otto 1999; Paquet and Robidoux 2001; Sbordone 1996). Their findings suggest that the SR with variable capital utilization represents true productivity shocks because it is invariant to demand shocks. The presumption is that procyclical movements in the SR represent a firm's response to market demand conditions rather than a true shift in its production function, and thus the SR overstates true productivity shocks. Therefore, the effects of cyclical demand shocks should be isohued from the SR if it is to represent technology shocks.

This line of reasoning gives rise to an important question regarding the properties of true technology shocks: Is the invariance hypothesis that technology shocks should be orthogonal to demand shocks a necessary or even sufficient condition, as suggested by many studies? This perception of technology shocks originates in empirical studies, which reported that variable utilization of capital greatly reduces the cyclicality of technology shocks. For example, Basu and Kimball (1997) reported that 40%-60% of SR cyclicality is caused by variable utilization, based on a model that includes cyclical utilization, increasing RTS, and variable technology simultaneously. They also indicated that variable utilization eliminates the evidence for increasing RTS, making price markup close to one, given the absence of large pure profits in the American economy. Paquet and Robidoux (2001) similarly demonstrate that the Canadian aggregate market structure is well described by constant RTS and perfect competition once capital stock is adjusted for variations in capacity utilization. These findings indicate that true technology shocks can be derivable by employing variable utilization of capital when estimating the residual. However, they also suggest that we can improve the measure for technology shocks by eliminating the effects of markup and increasing RTS from the SR, in addition to that of variable capital utilization, under conditions of increasing RTS and imperfect competition.

The main objective for removing cyclical movement from the SR is not to delete the correlation between and business cycles entirely, but simply to eliminate any error that may exist in the SR. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.