Economic data
It’s up, says one economist. Not really, say another; if you seasonally adjust it, it’s down.
They are probably talking about an economic time series. A time series is a series of data points, or measurements, of some aspect of the economy. These measurements are usually taken at regular intervals. Most time series that New Zealand economists look at are for quarterly data. We do have some monthly series, such as the data on retail sales, but it is often difficult to interpret monthly data given its volatility.
Volatility can be the result of a number of factors, including seasonality and sampling error. It pays to remind ourselves that most of our data comes via sample surveys rather than full censuses. And results from sample surveys have a sampling error associated with them. Most of the time, this sampling error will be small. Why do we rely on sample surveys? Simple, it costs so much less to survey a sample of, say, manufacturers than it does to survey every manufacturer in the country. But be aware that the sampling error is there, and we can never be entirely sure how big it is.
In looking at a time series, it pays to know what it’s actually measuring. This sounds like simple common sense. But how many people really know what the statement ‘business confidence is up’ really means? ‘Business confidence’ as measured by our own Quarterly Survey of Business Opinion is derived from a question which asks firms ‘Do you consider that the general business situation in New Zealand will improve, remain the same, or deteriorate over the next six months? So, as you can see, ‘business confidence’ relates to firms’ views on the New Zealand economy rather than to their views about their own activities. It often pays to go back to the original survey questionnaires in order to get a good understanding of what data means.
Another important feature of a data series is whether it relates to a flow or a stock. A flow is measured over a time period. For example, income is earned over a period of time. In contrast, wealth is a stock and is measured at a point in time.
Seasonality affects many time series in New Zealand. This isn’t surprising given that we have a large agricultural sector and output from this sector fluctuates markedly between winter and summer. There are statistical methods for correcting data series in order to take out regular seasonal patterns. The result is a series of seasonally adjusted values. Such series are important; if we are looking for economic turning points, where the economy changes direction, we are likely to spot these first by looking at seasonally adjusted data. A warning though: seasonal patterns are never totally regular, as a result and the seasonally adjusted series may only be a rough guide to the underlying changes in the series.
We often look at quarterly percent changes in seasonally adjusted data. It’s of little use looking at quarterly changes in unadjusted series that show strong seasonality – these quarterly changes will fluctuate violently. But we can look at annual percent changes unadjusted series, such as the change between the March quarter last year and the March quarter this year. Seasonality isn’t a problem here since we are comparing two quarters that are in the same part of the seasonal cycle. Economists also use annual average percent changes. An annual average is the average of four quarterly values. We compare this average with the average for the four quarters of the previous year to get the annual average percent change. In effect, it measures the year-on-year change, rather than the quarter-on-quarter change we get from the annual percent change referred to earlier.
Finally, we need to know the difference between nominal and real series. A nominal series is where values are simply expressed in dollars. A real series is where dollar values have been adjusted to account for inflation. This means they are expressed relative to a base period. For example, the series may by expressed in 1991/92 dollar values.
