PMIs: Reliable indicators for exports?

• DOI: http://doi.org/10.1111/roie.12395
• Date: 01 May 2019
• Published date: 01 May 2019
• Author: Sandra Hanslin Grossmann,Rolf Scheufele

Rev Int Econ. 2019;27:711–734. wileyonlinelibrary.com/journal/roie © 2019 John Wiley & Sons Ltd

|

711

1

|

INTRODUCTION

Assessing present and future economic conditions is crucial for decision‐makers in both the private

and the public sector. Exports are a key component of GDP and major driver of economic activity,

especially in small open economies. However, official information on exports is released with a time

lag, particularly for services. Early indicators such as domestic survey data on current sentiment and

industry conditions can provide valuable information on current export developments. The Purchasing

Managers’ Index (PMI) for the manufacturing sector is a well‐known example of a monthly survey

that provides timely indication of current and near‐term conditions in industry and has been found

to be useful in nowcasting and forecasting GDP changes (see, for instance, Koenig, 2002; Lahiri &

Monokroussos, 2013). However, in terms of exports, the PMIs of important trading partners could

provide even more valuable information, since they are a more direct indicator of foreign demand

conditions.

Received: 23 October 2015

|

Revised: 28 September 2018

|

Accepted: 9 November 2018

DOI: 10.1111/roie.12395

ORIGINAL ARTICLE

PMIs: Reliable indicators for exports?

SandraHanslin Grossmann

|

RolfScheufele

Swiss National Bank, Zurich, Switzerland

Correspondence

Rolf Scheufele, Swiss National Bank,

Economic Analysis, Börsenstrasse 15, P.O.

Box, Zurich 8022, Switzerland.

Email: rolf.scheufele@snb.ch

Abstract

Foreign economic activity is a major determinant of ex-

port developments. However, foreign GDP figures are

published too late to be useful for short‐term forecasting.

This paper presents a number of indicators based on the

widely available PMI surveys that provide very early sig-

nals of foreign activity. Using MIDAS models we analyze

the in‐ and out‐of‐sample performance of these and related

indicators for two very trade‐exposed countries (Germany

and Switzerland). We find that the monthly indicators

based on foreign PMIs are strongly correlated with quar-

terly export growth. The forecast comparison shows that

PMI‐based indicators perform very well relative to other

benchmark models.

JEL CLASSIFICATION

F14, F17, C53

712

|

GROSSMANN and SCHEUFELE

Survey‐based indicators such as the PMI are widely used by economic analysts for tracking the real

economy as they provide early signals of economic performance (Frale, Marcellino, Mazzi, & Proietti,

2010, 2011; Kaufmann & Scheufele, 2017). For instance, Koenig (2002) and Lahiri and Monokroussos

(2013) provide evidence of the usefulness of the PMI as an indicator for growth in the U.S. manufactur-

ing sector and the U.S. economy as a whole. Chudik, Grossman, and Pesaran (2016) also find that PMIs

are useful for output forecasts in an international context using the global vector autoregressive model

(GVAR) methodology, especially for nowcasting (but less for longer forecast horizons). Survey‐based

indicators, such as the (“Information und Forschung”—Institute for Economic Research) export cli-

mate indicator (Elstner, Grimme, & Haskamp, 2013) or foreign capacity utilization (Jannsen & Richter,

2012) have also been found to be successful short‐term indicators for exports.

In this paper we investigate the usefulness of foreign PMIs as a monthly indicator for nowcasting

and forecasting quarterly exports (both total exports and goods exports) as recorded in the national

accounts statistics. The major advantages of the PMIs are they are rapidly available on a monthly

basis, follow the same (or similar) definitions for all countries, and are only marginally revised. Our

investigation focuses on Switzerland and Germany, two European countries with a high share of total

exports in GDP (above 50%).1 As a consequence, these two economies are highly dependent on for-

eign activity and global trade. As such, the global trade collapse in 2008/2009, for instance, had a

direct, and huge impact on GDP in both countries, so early information on foreign activity would be

extremely valuable for getting a reliable view on the state of the Swiss and German economies.

We consider several variants of foreign PMI‐based indicators. We construct an indicator based

on export‐weighted manufacturing PMIs of the main trading partners of Switzerland and Germany.

Moreover, we examine three additional PMI‐based indicators: a factor extracted from the panel of for-

eign PMIs, the global manufacturing PMI calculated by J.P. Morgan, and the two domestic PMI indi-

cators (subindex “new exports” for Germany and the total PMI for Switzerland). Our analysis suggests

that PMI‐based indicators have valuable information for both Swiss and German quarterly exports.

Because of the different frequencies of the quarterly target variable and the monthly indicators,

we use mixed data sampling regression models for both the in‐sample and the out‐of‐sample analysis.

Our results indicate that the PMI information substantially improves the forecasting accuracy relative

to autoregressive benchmarks in all short‐term forecast settings (forecast, nowcast, and backcast). We

also examine the performance of the PMI‐based indicators relative to other export indicators, and find

that they also perform well compared with these additional benchmark models. Moreover, even when

no PMI data is available yet for the quarter of interest, PMI information from the previous quarter

helps to predict the development of quarterly exports in the current quarter. Our results show that an

indicator based on foreign PMIs outperforms other indicators and reduces the average forecast error

measure compared with an autoregressive benchmark substantially (up to 47% for Germany and up to

25% for Switzerland).

The remainder of the paper is organized as follows. In the next section, we discuss our data set and

the construction of the various export indicators based on foreign PMIs. In Section 3 we describe the

empirical methodology chosen to deal efficiently with the frequency mismatch between the quarterly

target variable and the monthly explanatory variables. Section 4 provides the in‐sample and out‐of‐

sample results, and Section 5 concludes.

2

|

DATA

Our focus is on quarterly real exports as reported in the national accounts as our dependent vari-

able (European System of Accounts 2010 standard). We use quarter‐on‐quarter growth rates of both