In the theory of finance, uncertainty plays a crucial role.
Economists often use the terms uncertainty and volatility
interchangeably, yet volatility is not the only form of uncertainty.
Firms face uncertainty about whether the economy is in an
expansionary or recessionary state, industries face regulatory
uncertainty, and individuals face uncertainty about risk premia. In
this dissertation, I consider the role that uncertainty about growth
rates, regulatory policy, and risk premia play in the investment
decisions of firms and individuals. The key theme linking the three
chapters is learning in dynamic environments.
In Chapter 1, I study the effects of demand growth uncertainty on
corporate investment decisions. In particular, how does uncertainty
about the state of the economy and the state of demand growth affect
a firm's decision to allocate capital to irreversible investment
projects? In the model, firms are able to choose both the timing
and scale of their investments, and the optimal scale will depend on
the unobserved state of demand growth. This second decision gives rise to an incentive to delay investment that does not exist in
standard real option models: When investment is irreversible, firms
risk allocating a sub-optimal level of capital to a project. Theoretically, I show how this incentive to delay is closely linked
to the benefits of learning about the economy. Empirically, using
estimated probabilities filtered from GDP growth, I find that 1)
beliefs about the economy inform corporate investment decisions, and 2) the relationship between investment and beliefs is quadratic.
In Chapter 2, I study an empirical extension of the model. Many
industries in the United States face regulatory uncertainty, and a
natural conjecture is that increased regulatory uncertainty has a
dampening effect on investment if 1) regulatory policy affects the
cash flows of the firm, 2) firms have flexibility over the scale of
their investments, and 3) regulatory uncertainty resolves quickly.
While regulatory uncertainty is not observable, I consider two
proxies: A variable indicating Presidential election years, and a
variable indicating divided government. The former is meant to
capture policy uncertainty associated with the possibility of a
change in government, while the latter is meant to capture policy
uncertainty associated with ideological variance. Empirically, both
measures are associated with a decrease in corporate investment
rates, consistent with the theoretical framework. The second
purpose of this chapter is to highlight the dangers of making
inferences about investment using inconsistent estimators and
regressions that fail to account for plausible alternative
hypotheses. Previous work linking investment to the political cycle
relies on least squares estimators that are inconsistent because the
firm-specific control variables are endogenous to the investment
decision. For a specific sub-sample of non-manufacturing firms, I
show that least squares estimates easily reject the null hypothesis,
while consistent first-difference estimates fail to do so. Finally,
I include a control for the fiscal environment of the federal
government, which helps to uncover important dynamics between
investment, the budget deficit, and the election cycle.
In chapter 3, I consider the currency hedging problem of a
risk-averse international investor who faces an unobservable
currency risk premium. A non-zero risk premium introduces a
speculative motive for holding foreign currency in the optimal
portfolio, and a time-varying risk premium introduces a
market-timing strategy. Uncertainty about the stochastic properties
of the risk premium significantly tames both the speculative and
market timing components, especially at long investment horizons,
and the optimal hedge approaches a complete hedge as risk aversion and the investment horizon increase. However, an investor who ignores the risk premium and fully hedges foreign investments faces a substantial opportunity cost because she forgoes the benefits of dynamic learning.