A Tort Liability Model of Accident Risk in Autonomous Vehicles
In Oh Cho, Chulyoung Kim
This study theoretically analyzes how the advancement of
autonomous driving technology affects automobile accident risk
using a tort liability framework. The analysis shows that under
strict liability, improvements in autonomous driving technology
may lead to increased driver inattentiveness, which can raise the
overall probability of accidents depending on the magnitude of this
behavioral response. In contrast, under the negligence rule, the
development of autonomous technology does not induce such
inattentiveness, and the accident risk consistently decreases.
Therefore, in contexts where public concern over autonomous
vehicle accidents is high, adopting a negligence-based regulatory
regime rather than strict liability may facilitate the faster and
broader adoption of autonomous driving technologies across
society.
Macroeconomic Forecasting Models using Search Indices
Sungdae Park, Hyun Hak Kim
Models for predicting macroeconomic variables such as economic
growth rate and inflation rate have traditionally relied solely on
structured data. In big data research, unstructured data such as
sentiment indices have also been introduced. In this study, we
introduced a forecasting model by considering the search volume
related to specific macroeconomic variables as indicative of the
variability of those variables. Meanwhile, for the search volume
index, we leveraged its rapid collection cycle to utilize the
characteristics of macroeconomic variables with different cycles.
We conducted both typicall forecasting and nowcasting using a
Bayesian model average mixture cycle forecasting model. Through
this model, we forecasted Korea¡¯s economic growth rate using
variables representing industrial production index, prices,
currency, oil prices, interest rates, labour, wages, and trade.
While search volume itself does not accurately reflect economic
information, it can reflect trends in public opinion or consumer
economic sentiment that existing macroeconomic data cannot
track, making it a useful supplementary variable for existing
economic models and potentially improving forecasting power.
Even in a very simple form of macroeconomic forecasting model,
the search results index demonstrates improved forecasting power,
and we expect that expanding this to general equilibrium models
could yield significant benefits.