|dc.description.abstract||This thesis provides contemporary insight into the safe haven phenomenon. We separately examine
three characteristic periods; the years 2001 to 2007, 2007 to 2010, and 2010 to 2016. Our focus is
to examine currency portfolio rebalancing in times of increased risk aversion and identify any
periodical changes in behavior pre and post financial crisis.
Using an autoregressive distributed lag (ADL) model, we study the high-frequency movements of
eight nominal effective exchange rates against three measures of risk aversion. The purpose of the
ADL model is to examine the safe haven behavior on an average basis. We later expand this
baseline model to an interactive dummy model, which allows us to explore the more conditional
behavior during crisis episodes. To the best of our knowledge, the literature has yet to explore this
topic in a similar fashion, especially for the recent years 2010 to 2016.
We document that the Japanese Yen (JPY), Swiss Franc (CHF) and U.S. Dollar (USD) tend to
appreciate when there is an increase in i) stock volatility; ii) forex volatility; iii) composite financial
volatility. In recent years, the JPY shows significantly stronger safe haven tendencies, whereas the
CHF portrays weaker properties post financial crisis. The USD has experienced a noteworthy shift
in status, and shows strong signs of being a safe haven currency post financial crisis compared to
the years 2001 to 2007, where it behaved more pro-cyclically with financial markets.
The New Zealand Dollar (NZD) and Australian Dollar (AUD) tend to depreciate when risk
aversion increases. Here, the AUD shows stronger non safe haven tendencies than the NZD.
Interestingly, the Norwegian Krone (NOK) shows relatively stronger non safe haven tendencies
for the recent period 2010 to 2016. On the other hand, results for the Euro (EUR) and British Pound
(GBP) are overall inconclusive.
We also identify a tendency of stronger quantitative impacts of the JPY during risk episodes
compared to ordinary days. The safe haven phenomenon is however not contingent upon these
specific episodes. On average, the quantitative impacts and explanatory powers are at their highest
for all findings during the years of the financial crisis, 2007 to 2010. Furthermore, the years 2010
to 2016 show far more powerful safe haven flows than 2001 to 2007.||nb_NO