The report departs from conventional time series analysis and investigates the existence of long memory (LRD) in the stream of daily visitors, arriving from various sources to New Zealand from 1997 to 2010, using selected estimators of the Hurst-exponent. The daily arrivals of visitors are treated as a stream of "digital signals" with the inherent noise. After minimizing the noise (i.e. the presence of short-term trends, periodicities, and cycles) we found the existence of significant long memory embedded in our data of daily visitors from all sources and in the aggregate. Strong evidence of embedded “long memory” implies that Joseph Effect – that good times beget good times and bad times beget bad – whose existence in the underlying process may have interesting implications for tourism policy makers. Our findings suggest evidence of such long term memory in tourist arrival data. Further, unless this long memory effect is taken into consideration, any traditional statistical analysis based on Gaussian and Poisson assumptions may be overly biased.
