3  A composite index of leading indicators of New Zealand employment

The first step in developing a composite index of leading indicators is to identify potential component series. When constructing a composite index, it is useful to start on a broad basis with a large number of variables and then drop variables that do not add information. This way no potential information is excluded. Composite indexes do not attempt to explicitly identify the sources of variation in the reference series. Rather the approach focuses on finding variables that tend to be affected by the same shocks as the reference series but earlier than the reference series.

3.1  Selection of initial series

To construct a composite index of leading indicators of New Zealand employment (CI) we started with about 140 series that broadly cover the different sectors of the New Zealand economy.[3] The first step in the analysis consisted of graphing the data with employment and calculating some descriptive statistics, such as correlation coefficients.

The 140 or so series that we considered can be grouped into eight main categories: (i) aggregate economic activity, (ii) consumption, (iii) investment, (iv) trade, (v) financial and monetary variables, (vi) business and consumer confidence, (vii) labour market indicators, and (viii) foreign variables. These indicators were thought to lead New Zealand employment for the following reasons.

Aggregate economic activity, consumption and investment

Employment growth generally lags output growth and a rise or fall in domestic demand or its components should give some indication about future employment growth. The housing market has been particularly important in New Zealand. Following the deregulation of domestic financial markets in the mid-1980s, households gained access to finance previously not available. This led to an increase in household borrowing and strong gains in the housing sector. Moreover, during the mid-1990s an increase in permanent long-term migration led to demand pressures and increased demand for housing, which in turn contributed to an upswing in employment.

Trade indicators

Since the economic reforms in the mid-1980s and early 1990s, New Zealand has undergone significant trade liberalisation and become one of the most open countries in the OECD. The share of exports in output has increased from about 25 percent in 1985 to around 35 percent currently and the export sector is a growing employer.

Financial and monetary variables

Financial and monetary indicators, such as interest and exchange rates, interest spreads and the money supply, are potential leading indicators of employment as they capture the effects of monetary policy on economic activity. Until the mid-1980s, consumer price inflation was at double-digit rates except for the price, rent and wage freezes in 1983-84. Dissatisfaction with the systematic under-performance of the New Zealand economy compared to other countries during the 1970s and 1980s prompted the government of the day to embark on a period of comprehensive economic reforms. Macroeconomic stability through fiscal restraint and monetary policy focused on reducing inflation was one of the key themes, leading to a period of high and rising real interest rates.

Consumer and business confidence indicators

Following the structural and fiscal adjustments, disinflation and a collapse in the New Zealand sharemarket, the rate of economic growth in New Zealand was poor and unemployment high. However, since the end of the recession in the early 1990s, New Zealand’s economic performance has been improving and growth was particularly strong during the mid-1990s. Consumer and business confidence were likely important contributors to these swings in economic activity.

Labour market indicators

Labour market indicators, including hours worked or employment intentions, are direct leading indicators of future employment. Typically, hours worked, for example, rise prior to increases in total employment. This is because it is easier for employers to have employees work over-time than to hire additional staff. As most workers are only willing to work over-time for a limited period of time, new staff are generally employed if the workload remains high for a prolonged period of time. So, growth in hours worked often leads employment changes.

Foreign variables

As New Zealand is a small open economy heavily reliant on exports, foreign activity is an important contributor to economic performance. Strong growth in its trading partners ultimately translates into more jobs in New Zealand. Historically, New Zealand’s growth was importantly influenced by economic activity in the United Kingdom. However, since the late 1980s and 1990s trade has shifted towards the economies of Australia, the United States, Japan and other Asian countries, and New Zealand has benefited from lengthy economic expansions in some of these countries, in particular the United States.

3.2  The final choice of variables and their weights

The final choice of variables to include in the composite index of New Zealand employment was based on their concordance with employment. The concordance statistic, which measures the proportion of time employment and a leading indicator move in the same direction, is calculated as follows

(8)

where is a series equal to one when the change in employment is zero or positive and zero when the change in employment is negative. The series for the indicator series is defined accordingly and T denotes the sample size for which both series are available.[4] As a proportion, the values that can take are bounded between zero and one.

The concordance statistic was used instead of correlation coefficients as correlation coefficients mix amplitude and duration measures, and the amplitude of a particularly large swing that is common to two variables may dominate the covariance of the two series. The concordance statistic avoids this problem.

Concordance statistics were calculated for employment and lags of all potential component series. Lagged concordance was considered as the objective is to build a composite index of leading indicators (rather than of coincident indicators).[5] Series with the largest degree of concordance, lagged one quarter or more, were then included in the composite index. The weight of each component series was determined as follows

(9)

for . For example, if last quarter job ads, move in the same direction as this period employment 80 percent of the time, then their weight (in equation 5) is 0.8 divided by the sum of (non-normalised) weights. Moreover, if job ads have a large weight in the composite index, then the composite index is likely to lead employment by one quarter as the concordance is largest between employment and job ads lagged one quarter.

Alternatively, we could have used equal weights for all component series. However, the restriction of equal weights decreased the concordance (and correlation) between the composite index of leading indicators and employment, and worsened the forecasting performance of the indicator model.