Dispersed Beliefs, Frictional Acquisition of Capital, and Investment Dynamics

I show both empirically and theoretically that dispersion of firms' beliefs about the latent aggregate productivity, measure of "aggregate ambivalence," is a distinctively stronger driver of investment dynamics as compared to the volatility-based measures of fundamental uncertainty. This paper answers the question about how firm's learning with noises affects micro-level and aggregate investment over the business cycles. I document that increasing disagreement among firms not only reduces aggregate investment but has non-linear amplifying effect conditional on the total factor productivity. I then develop a DSGE model with second moment shocks to both firms' ex-ante beliefs and fundamental productivity, endogenous partial investment irreversibility, and firms' imperfect learning. Irreversibility is endogenously determined in a market of used capital with search-match friction that captures the idea that used capital can bring about restructuring benefit. As investing firms accumulate capital as investment through acquiring used capital from disinvesting firms and combining them with the new capital goods. This matching process is frictional as it incurs a cost of leaving the acquisition deal unsettled, a negotiation cost that increases in the firms’ disagreement. Greater "ambivalence" thus induces larger fraction of used capital liquidizers and increases the acquirers' information cost of holding unfulfilled acquisitions. Hence larger belief dispersion increases transaction cost of capital resale and binds aggregate irreversibility constraint, which generates the linear effect of aggregate ambivalence. Non-linear response of investment to changes in belief dispersion is a result of learning procedure via misattribution of signals. While increasing "ambivalence" in bad times reduces aggregate investment but increases it in good times. Counter-cyclical information contain policy may be welfare improving.