How many jobs? A leading indicator model of New Zealand employment

Claus, Edda; Claus, Iris Working Paper How many jobs? A leading indicator model of New Zealand employment New Zealand Treasury Working Paper, No. 02/13 Provided in Cooperation with: The Treasury, New Zealand Government Suggested Citation: Claus, Edda; Claus, Iris (2002) : How many jobs? A leading indicator model of New Zealand employment, New Zealand Treasury Working Paper, No. 02/13, New Zealand Government, The Treasury, Wellington This Version is available at: https://hdl.handle.net/10419/205488 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ How many jobs? A leading indicator model of New Zealand employment Edda Claus and Iris Claus N EW Z EALAND T REASURY W ORKING P APER 02/13 J UNE 2002 NZ TREASURY WORKING PAPER SERIES 02/13 How many jobs? A leading indicator model of New Zealand employment MONTH / YEAR June/2002 AUTHORS Edda Claus The Treasury Langton Cresent Canberra ACT 2600 AUSTRALIA Email Telephone Fax [email protected] 61-2-6263 2808 61-2-6263 2734 Iris Claus The Treasury 1 The Terrace PO Box 3724 Wellington NEW ZEALAND Email Telephone Fax [email protected] 64-4-471 5221 64-4-499 0992 ACKNOWLEDGEMENTS We would like to thank Richard Downing, Leo Krippner and Robert Lamy for valuable comments and Joanne Archibald for helpful suggestions about the data. The paper has also benefited from comments by participants of a seminar at the Reserve Bank of New Zealand and the 2001 Australasian Macroeconomic Workshop in Adelaide and two anonymous referees. NZ TREASURY New Zealand Treasury, PO Box 3724, Wellington 6008, NEW ZEALAND Email Telephone Website [email protected] 64-4-472 2733 www.treasury.govt.nz DISCLAIMER The views expressed in this Working Paper are those of the author(s) and do not necessarily reflect the views of the New Zealand Treasury. The paper is presented not as policy advice, but with a view to inform and stimulate wider debate. WP02/13 How many jobs? A leading indicator model of New Zealand employment i Abstract This paper constructs a composite index of leading indicators of New Zealand employment. The choice of variables and their weights in the composite index are determined by their concordance with employment. The composite index is included in an indicator model to forecast quarterly employment growth. The indicator model explains about 67 percent of the quarterly variation in employment, in sample, and correctly predicted the direction of next period employment almost 80 percent of the time, out of sample. JEL CLASSIFICATION C53 (forecasting, mathematical and quantitative methods), E37 (forecasting, macroeconomics) KEYWORDS Composite index of leading indicators, forecasting, employment WP02/13 How many jobs? A leading indicator model of New Zealand employment ii Contents Abstract.................................................................................................................................i Contents...............................................................................................................................ii List of Tables .............................................................................................................ii List of Figures .............................................................................................................ii 1 Introduction.................................................................................................................1 2 Constructing a composite index of leading indicators.................................................2 2.1 Standardisation and weighting of individual component series ..................................... 2 2.2 Standardisation and cumulation of the composite value ............................................... 3 3 A composite index of leading indicators of New Zealand employment.......................4 3.1 Selection of initial series................................................................................................. 4 3.2 The final choice of variables and their weights .............................................................. 5 3.3 The final composite index of leading indicators ............................................................. 6 4 A forecasting model of New Zealand employment growth .........................................9 5 Forecast performance relative to two benchmark models ........................................11 6 Concluding remarks..................................................................................................13 Appendix A: Series considered in the construction of the composite index of leading indicators.......................................................................................................15 Appendix B: Data and data sources ..................................................................................19 References.........................................................................................................................21 List of Tables Table 1 - Weights of the components in the composite index of leading indicators............ 7 Table 2 - Estimation results .............................................................................................. 10 Table 3 - Relative performance of the indicator model ..................................................... 12 Table 4 - Confusion matrices ............................................................................................ 13 List of Figures Figure 1 - Employment and the composite index................................................................ 8 Figure 2 - Change in employment and the composite index............................................... 9 Figure 3 - Actual and fitted employment growth ............................................................... 11 Figure 4 - Out-of-sample forecasts of employment growth............................................... 12 WP02/13 How many jobs? A leading indicator model of New Zealand employment 1 How many jobs? A leading indicator model of New Zealand employment 1 Introduction Forecasts play a key role in formulating fiscal and monetary policy in New Zealand. Under the Fiscal Responsibility Act 1994, which sets out the principles of responsible fiscal management, The Treasury is required to prepare forecasts of the New Zealand economy that incorporate the impact of the short-term fiscal intentions and long-term fiscal objectives of the government of the day. The Reserve Bank of New Zealand Act 1989 provides the legislative framework for conducting monetary policy and requires the Reserve Bank of New Zealand to publish assessments of the New Zealand economy, which provide the basis for setting policy to maintain inflation within the target band. The main tool used by The Treasury to produce medium-term projections is a large structural macroeconomic model, the New Zealand Treasury Model (NZTM), and for the Reserve Bank of New Zealand the Forecasting and Policy System (FPS).1 However, large-scale structural models are generally not well suited to making short-term assessments as they focus on the medium-term dynamics of the economy. To help evaluate short-term fluctuations and cyclical turning points, statistical techniques and econometric forecasting tools, such as indicator models, are used instead. The purpose of this paper is to develop a composite index of leading indicators of New Zealand employment that can be used to forecast quarterly employment growth. One advantage of composite indexes is that they allow incorporating information from many variables that could not be included in time series forecasts with short sample periods. Moreover, composite indexes may be less subject to instability due to policy changes. For example, the composite index of leading indicators of New Zealand employment includes firms’ employment intentions. It is hard to imagine any government policy that could substantially or permanently alter the relationship between employment intentions and actual job growth. The remainder of the paper is organised as follows. Section 2 outlines the construction of a composite index of leading indicators. Section 3 develops such an index for New Zealand employment. The choice of component variables and their weights in the 1 NZTM is described in (Szeto 2002). A discussion of FPS can be found in (Hunt, Rose and Scott 2000). WP02/13 How many jobs? A leading indicator model of New Zealand employment 2 composite index are determined by their concordance with employment. The composite index of leading indicators is then included, as an explanatory variable, in a singleequation model to forecast quarterly employment growth. The indicator model is described in section 4 and its out-of-sample forecasting performance is evaluated relative to two benchmark models in section 5. Conclusions are presented in section 6. 2 Constructing a composite index of leading indicators 2 The idea of using leading economic indicators in business cycle forecasting was originally developed by (Mitchell and Burns 1938) at the National Bureau of Economic Research (NBER). A composite index of leading indicators is the (generally) weighted average of several component series. Composite indexes are constructed because they tend to be more reliable as a cyclical indicator than any of its components taken individually. This is partly because much of the independent measurement error and other noise in the component series is smoothed out in an index. The algebraic construction of a composite index of leading indicators involves two main steps: (i) standardisation and weighting of the individual component series, and (ii) standardisation and cumulation of the composite value. Both steps are explained in more detail in this section, while the selection of variables to include in an index of New Zealand employment is discussed in section 3. 2.1 Standardisation and weighting of individual component series The first step in the construction of a (quarterly) composite index of leading indicators is to calculate quarter-to-quarter symmetrical percentage changes, t Y , for each individual component j of the composite index j 1t j t j 1t j t j tXX XX 200Y − − + − ∗= (1) where j t X denotes component j at time t. For series that contain zero or negative values, or that are already in index or percentage form, the following formula is used instead )XX(200Y j 1t j t j t− −∗= (2) For series that are a difference of two series (such as the interest spread) or a ratio, equation (3) is used j t j tXY = (3) 2 The section partly follows (Zarnowitz and Boschan 1975). See also (Green and Beckman 1993). WP02/13 How many jobs? A leading indicator model of New Zealand employment 3 The next step is to standardise the transformed series to prevent a volatile component from dominating changes in the composite index. Each transformed component variable is standardised by dividing it by its historical average without regard to sign, i.e., ∑ − =j t j t j t Y 1T 1 Y S (4) where T denotes the number of observations. The transformed and standardised series, j t S , are then combined into a composite variable using equation (5) ∑ =j t j tS*wI (5) where t I is the raw composite value and j w the individual weight of each series normalised to sum to one. One way to aggregate the components j t S is to assume equal weights. However, a number of alternative weighting schemes are available to better reflect the relationship and the importance of each component with the reference series, employment in our case. (Choosing the weights is discussed further in the next section.) 2.2 Standardisation and cumulation of the composite value The purpose of the second step, standardisation and cumulation of the composite value, is to transform the raw composite value t I so that it has the same historical average (without regard to the sign) as the reference series. The standardised composite value, s t I , is obtained as ∑ ∑ = |R| |I| I I j t j t t s t (6) where j t R is the symmetrical percentage change of the reference series (employment). The standardised composite value s t I is then transformed into an index using the formula in equation (7) )I200( )I200(*I CI s t s t s1t t− + =− (7) where t CI is the standardised composite index. WP02/13 How many jobs? A leading indicator model of New Zealand employment 4 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 3 A complete list of the variables is included in Appendix A. WP02/13 How many jobs? A leading indicator model of New Zealand employment 11 Figure 3 - Actual and fitted employment growth 5 Forecast performance relative to two benchmark models To assess the out-of-sample forecast performance of the indicator model, we estimated two benchmark models for comparison: (i) a univariate ARIMA (2,1,3) model of employment growth and (ii) a two-variable vector autoregression (VAR) model including employment growth and changes in the composite index of leading indicators.10 To evaluate the forecast performance of the indicator model and the two benchmark models, we generated out-of-sample forecasts using a fixed length, rolling window, time varying coefficient approach for the period 1996Q1 to 2001Q4. With this technique, we first estimated all three models over the period 1987Q2 to 1995Q4. The estimated coefficients from each model were used to forecast employment growth one quarter ahead. The models were then rolled forward one quarter and re-estimated. We repeated this process until the last observation was reached. This led to 24 one-quarter-ahead outof-sample forecasts. Out-of-sample forecasts of employment growth from the indicator model and the two benchmark models, plotted in Figure 4, all appear to follow actual employment growth quite closely. To compare the forecast performance of the three models more formally, we calculated the mean absolute forecast error (MAE), the root mean squared forecast error (RMSE) and the U-Theil for each model. The smaller the values of the MAE and the RMSE are, the more accurate, on average, the forecast of a model. The smaller the value 10 A lag length of five was determined by minimising the Schwarz information criterion. -80 -60 -40 -20 0 20 40 1987 1989 1991 1993 1995 1997 1999 2001 thousands -20 -10 0 10 20 30 40 thousands Error (RHS) Actual (LHS) Fitted (LHS) WP02/13 How many jobs? A leading indicator model of New Zealand employment 12 of the U-Theil, the better the model performs compared to a naïve forecast of no change.11 The results are reported in Table 3. Figure 4 - Out-of-sample forecasts of employment growth Table 3 shows that the indicator model outperforms both the VAR and ARIMA models in terms of out-of-sample forecast ability. All three evaluation criteria, the MAE, the RMSE and the U-Theil statistic, are lower for the indicator model than for the two benchmark models. The VAR model outperforms the ARIMA model in terms of the RMSE and UTheil. The composite index thus appears to be particularly useful in forecasting employment changes. Table 3 - Relative performance of the indicator model MAE * RMSE * U-Theil indicator model 6.238 7.744 0.693 ARIMA (2, 1, 3) 6.967 9.310 0.834 VAR model 7.464 8.900 0.797 * in thousands 11 The U-Theil is calculated as the root mean squared forecast error of the model divided by the root mean squared forecast error of the naïve model of no change. -20 -10 0 10 20 30 40 1996Q1 1997Q1 1998Q1 1999Q1 2000Q1 2001Q1 thousands -20 -10 0 10 20 30 40 thousands Actual employment Indicator model ARIMA VAR WP02/13 How many jobs? A leading indicator model of New Zealand employment 13 Finally, Table 4 reports the 2x2 confusion matrices for the three models. The confusion matrix records the number of times a model correctly predicted the direction of next period employment growth out of sample.12 The upper diagonal (upper-left) element records the number of times a model correctly predicted an increase in employment growth, while the lower diagonal (lower-right) element reports how often the model correctly forecast a decrease. For example, the indicator model correctly predicted fourteen rises and five declines in employment growth. The off-diagonals report the number of times a model missed the direction of employment changes. The lower-left (upper-right) off-diagonal element records the number of actual moves in employment growth that were up (down) while the predicted changes were down (up). Table 4 - Confusion matrices actual outcome indicator model ARIMA (2, 1, 3) VAR 14 4 12 6 14 4 model prediction 1 5 3 3 1 5 The results in Table 4 suggest that the indicator model and the VAR model forecast the direction of next period employment growth reasonably well. The indicator and VAR models have five false signals each, which is better than the ARIMA with nine false signals. The indicator and VAR models correctly forecast the direction of employment almost 80 percent of the time compared to about 63 percent for the ARIMA. Both the indicator and VAR models predicted employment growth to increase when it actually fell more often than they predicted employment growth to fall when it actually increased. The out-of-sample forecast performance of the indicator and VAR models suggests that the composite index of leading indicators is a good predictor of employment changes. 6 Concluding remarks In this paper, we constructed a composite index of leading indicators of New Zealand employment that was then used in an indicator model to forecast quarterly changes in employment. The indicator model is able to explain about 67 percent of the quarterly variation in employment, in sample, and correctly predicted the direction of next period employment almost 80 percent of the time, out of sample. The composite index of leading indicators thus appears to be a good predictor of employment changes. 12 See (Paquet, Fauvel and Zimmermann 1999). WP02/13 How many jobs? A leading indicator model of New Zealand employment 14 A VAR model including changes in the composite index and employment produces results only slightly worse than the indicator model. Thus, to produce out-of-sample forecast for more than one quarter ahead, the indicator model, which can be interpreted as a restricted VAR, could be re-estimated as a VAR (using seemingly unrelated regression estimation). The estimation of the indicator model as a restricted VAR is left for future work. WP02/13 How many jobs? A leading indicator model of New Zealand employment 15 Appendix A: Series considered in the construction of the composite index of leading indicators 13 13 All data are seasonally adjusted apart for interest and exchange rates and price indexes. Domestic aggregate economic activity real production-based gross domestic product (GDP) real expenditure-based GDP real production-based GDP: manufacturing real production-based GDP: construction real production-based GDP: wholesale trade real production-based GDP: transport and storage real production-based GDP: communications real production-based GDP: services real production-based GDP: change in total stocks Consumption private consumption (real) private consumption of housing (real) private consumption of durables (real) private consumption of non-durables (real) net tourist consumption (real) total household consumption (real) electricity generation: sales to custumers retail sales excluding automotive and personal services Investment new dwelling consents total dwellings REINZ number of dwellling sales market investment (real) market investment in plant and machinery (real) non-residential market investment (real) total investment in plant and machinery (real) non-residential total investment (real) Trade merchandise imports merchandise exports exports (real) exports of goods (real) volume of merchandise exports volume of merchandise imports WP02/13 How many jobs? A leading indicator model of New Zealand employment 16 exports of meat and meat products (real) livestock slaughter dairy exports (real) arrivals, overseas visitors Financial and monetary variables consumer price index adjusted for the introduction / increase in the goods and services tax (CPI) food price index monetary conditions index (MCI) MCI divided by CPI 1 year government bond yield 2 year government bond yield 5 year government bond yield 10 year government bond yield 30 day bank bill yield 60 day bank bill yield 90 day bank bill yield 10 year government bond yield minus 1 year bond government bond yield 10 year government bond yield minus 2 year bond government bond yield 10 year government bond yield minus 5 year bond government bond yield 5 year government bond yield minus 1 year bond government bond yield 5 year government bond yield minus 2 year bond government bond yield 10 year government bond yield minus 30 day bank bill yield 10 year government bond yield minus 60 day bank bill yield 10 year government bond yield minus 90 day bank bill yield 5 year government bond yield minus 30 day bank bill yield 5 year government bond yield minus 60 day bank bill yield 5 year government bond yield minus 90 day bank bill yield call rate NZD/AUD exchange rate: average 11am NZD/USD exchange rate: average 11am trade weighted index (TWI): average 11am NZD / AUD exchange rate: end of period NZD / USD exchange rate: end of period TWI with end of period bilateral exchange rates total billings on New Zealand credit cards total advances on credit cards outstanding notes and coins held by the public WP02/13 How many jobs? A leading indicator model of New Zealand employment 17 M1 M1 divided by CPI M2 M2 divided by CPI M3 M3 divided by CPI resident M3 resident M3 divided by CPI private sector credit resident private sector credit domestic credit resident domestic credit New Zealand stock exchange capital total: price index (NZSE) NZSE divided by CPI Business and consumer confidence Quarterly Survey of Business Opinion (QSBO): general business situation QSBO: find skilled labour QSBO: find unskilled labour QSBO: limiting factor -- labour QSBO: limiting factor -- capital QSBO: limiting factor -- other QSBO: new investment in buildings QSBO: new investment in plant and machinery QSBO: number of employees (past three months) QSBO: overtime worked (past three months) QSBO: labour turnover (past three months) QSBO: average costs (past three months) QSBO: average selling price (past three months) QSBO: profitability (past three months) QSBO: number of employees (next three months) QSBO: overtime worked (next three months) QSBO: labour turnover (next three months) QSBO: average costs (next three months) QSBO: profitability (next three months) QSBO: economy-wide capacity utilisation National Bank Business Outlook (NBBO): commercial construction intentions (next 12 months) NBBO: residential construction intentions (next 12 months) WP02/13 How many jobs? A leading indicator model of New Zealand employment 18 NBBO: employment expections, total all sectors (next 12 months) NBBO: business confidence, total all sectors (next 12 months) NBBO: activity outlook, total all sectors (next 12 months) NBBO: investment intentions, total all sectors (next 12 months) Wholesale Trade Survey (WTS): sales WTS: stocks WTS: stock to sales ratio consumer confidence, one network news, Colmar Brunton poll Labour market indicators Household Labour Force Survey (HLFS) hours worked HLFS: unemployment rate HLFS: participation rate A NZ job advertisement permanent and long-term migration, net actuals external migration, total arrivals long-term migration, arrivals Quarterly Employment Survey: total paid hours, all industries average ordinary time wages divided by CPI Foreign variables A NZ commodity price inde x US industrial industrial production, total index Conference Board US composite leading indicator Conference Board US coincident leading indicato r National Association of Purchasing Management (NAPM): purchasing managers' index NAPM: production index NAPM: new orders index NAPM: deliveries index NAPM: inventories index NAPM: employment index NAPM: commodity price index NAPM: imports index NAPM: new export orders index NAPM: US help wanted index 3 months US Treasury notes 10 year US Treasury notes 10 year US Treasury notes minus three months US Treasury notes US M1 divided by US consumer price index WP02/13 How many jobs? A leading indicator model of New Zealand employment 19 Appendix B: Data and data sources employment HLFS official employed. Thousands. Seasonally adjusted. Source: Statistics New Zealand. retail Retail sales excluding automotives and personal services. New Zealand dollar million. Seasonally adjusted. Source: Statistics New Zealand and Reserve Bank of New Zealand. consent New dwelling consents. New Zealand dollar million. New Zealand dollars. Seasonally adjusted. Source: Statistics New Zealand. inventories Change in total stocks. Constant New Zealand dollar million. Seasonally adjusted. Source: Statistics New Zealand. arrivals Short-term migration: arrivals, overseas visitors. Thousands. Seasonally adjusted. Source: Statistics New Zealand. spread 5 year government bond yield minus 30 day bank bill yield. Percent. Source: Reserve Bank of New Zealand. TWI Trade weighted index with end of period bilateral exchange rates. Source: Reserve Bank of New Zealand. intentions Number of employees, next three months. Index. Seasonally adjusted. Source: New Zealand Institute of Economic Research Quarterly Survey of Business Opinion. ads ANZ job advertisements, three main centres. Thousands. Seasonally adjusted. Source: ANZ. Employment (in thousands) 1400 1450 1500 1550 1600 1650 1700 1750 1800 1850 1900 1985 1988 1991 1994 1997 2000 WP02/13 How many jobs? A leading indicator model of New Zealand employment 20 Intentions (index) -50 -40 -30 -20 -10 0 10 20 30 1985 1988 1991 1994 1997 2000 Inventories (in constant NZ$M) -600 -400 -200 0 200 400 600 800 1985 1988 1991 1994 1997 2000 Spread (in percent) -10 -8 -6 -4 -2 0 2 4 1985 1988 1991 1994 1997 2000 Arrivals (in thousands) 100 150 200 250 300 350 400 450 500 550 1985 1988 1991 1994 1997 2000 New dwelling consent (in NZ$M) 100 150 200 250 300 350 400 1985 1988 1991 1994 1997 2000 TWI (index) 40 45 50 55 60 65 70 75 80 1985 1988 1991 1994 1997 2000 Job advertisement (in thousands) 5 10 15 20 25 30 1985 1988 1991 1994 1997 2000 Retail sales (in NZ$M) 3000 4000 5000 6000 7000 8000 9000 1985 1988 1991 1994 1997 2000