Analysis of Business Activity Indicators in Russia

Размещено на http://www.allbest.ru/

ФЕДЕРАЛЬНОЕ ГОСУДАРСТВЕННОЕ АВТОНОМНОЕ ОБРАЗОВАТЕЛЬНОЕ УЧРЕЖДЕНИЕВЫСШЕГО ОБРАЗОВАНИЯ

«НАЦИОНАЛЬНЫЙ ИССЛЕДОВАТЕЛЬСКИЙ УНИВЕРСИТЕТ

«ВЫСШАЯ ШКОЛА ЭКОНОМИКИ»

Международный институт экономики и финансов

Выпускная квалификационная работа -

БАКАЛАВРСКАЯ РАБОТА

по направлению подготовки 38.03.01 «Экономика»

образовательная программа «Программа двух дипломов по экономике НИУ ВШЭ и Лондонского университета»

Analysis of Business Activity Indicators in Russia

Турбина Мария Александровна

Научный руководитель

ученая степень, звание

Н.Г. Каныгина

Introduction

Opportunity to easily and quickly estimate current and forthcoming changes in the economic activity is vital for all the economic agents. Due to the fact that official macro statistics is always published with a substantial delay, business activity indicators have been established, which allow its users to promptly get updates on current business climate and its expectations, as these indices are usually posted at monthly frequency. A major amount of organizations, business men and central banks all over the world are checking information, which these indicators of business activity demonstrate, to get a general insight of what is happening with economic activity, what are the sentiments of business community and their expectations about nearest future. On the basis of this data, economic agents make important business choices. Therefore, financial organizations and policy institutions all across the world try to construct their own business activity indicators, and for that purpose they use business tendency surveys. Hence, these indicators reflect opinion of the agents about the current situation in the economy and their predictions about its nearest future. So, it is crucial for any country to understand how well its currently used business activity indicators are working, whether they depict all the necessary information for the economic actors.

Business tendency indicators are considered as subjective, but promptly available and simple in interpretation measures of business confidence, that are closely related to economic variables, such as GDP, and that are capable of prediction of changes in the economic activity. So, all over the world various business activity indicators are considered to be the reliable leading indicators, which means that they are deemed to foresee the forthcoming direction of economic activity from a few months ahead, and especially, recessions. So, in most of the research papers, relationship of a business tendency indicator with economic activity is tested, as well as usefulness of such an indicator in identification of fluctuations in this activity. However, there is a scarcity of research on that topic regarding Russia, as application of various business activity indicators is not long, it is just around 25 years. So, there is no information at all about which of the implemented business activity indices are efficient and credible. Hence, a thorough investigation of these aspects is crucial. That is why, a goal of this paper is to provide theoretical discussion, as well as empirical analysis of these indices, which are, namely, the Purchasing Managers' Index, the EntrepreneurialConfidence Index, the Business Confidence Index and the Russia Small Business Index. In particular, their association with economic activitywill be testedand then, their power to detect and predict direction of this activity will be verified. This analysis will help to reveal which indicators are indeed associated with economic activity and hence, credible and useful for decision-making.

Results of the given paper highlight that not all of the business activity indicators that are implemented in Russia are useful and provide credible data about economic activity. Such an unreliable indicator is appeared to be the BCI, whose correlation behaviour has revealed vary volatile and unpredictable relationship with economic growth, and then its regression testing has confirmed that this index is not significant. Speaking about indices that have showed good results, the PMI and ECI are both turned out to be quite closely related to economic growth, and then, they demonstrated pretty accurate detection of changes in this variable, what makes them trustworthy indicators. Finally, in regards to a new index, the RSBI, there is a good prediction quality of economic growth, but correlation analysis has pointed out confusing results. In general, in contrast to a situation in the many countries, business tendency indices in Russian economy exhibit very volatile correlation with economic growth, as this correlation changes quite frequently from positive values to the opposite ones. This may be a signal that these indicators miss some important information to be incorporated in their structure, which could have been able to explain such a behaviour.

This paper is organised as follows, firstly, methodology and history of each business activity indicator will be discussed to a full extent. Then, a literature review will cover the information about the theoretical and practical studies of application of these indices for various economies. Then, cross-correlation and rolling correlation for each of the indicators will be performed to assess degree of relationship of them with economic growth. Later, regression analysis with the help of logit model will be carried out to detect which indicators are significant, and moreover, which of them has a good prediction power in identification of changes of economic activity.

business economic relationship

1. Methodology of business activity indices

1.1 General structure of business activity indices

As all the business confidence indicators are based only on managers' opinion and perception, they are all subjective indices. On the one hand, using of indicators based on opinions involves many positive aspects and advantages over the usage of the official macro variables and indices. Among the benefits are the fact that they are reliable in terms of representation of the overall business conditions, and also, that such indicators are obtained and being published pretty quickly (within 1 month), and their results are not changed ever afterward (Ozyldirim et al., 2011). Moreover, these indicators serve as predictors of fluctuations in many economic variables, such as real GDP, industrial production, inventories, sales, commodity prices (Ozyldirim et al, 2011). They depict economic climate of the main private sectors - such as manufacturing and services. And in addition to that, methodology of international business confidence indicators is such that it is universal in terms of calculation among the different countries, hence, one can easily compare the economic conditions and expectations of various countries based on one confidence index (whether it be PMI, BCI or some other index) (Bose, 2015). It is useful, because official macroeconomic variables (for instance, such as GDP) can be calculated very differently across the world, so their figures cannot be compared directly.

On the other hand, methodology of such indices inevitably implies some drawbacks. The main disadvantage is considered to be the fact that they do not represent the depth of the future changes in the variables (Koenig, 2002). Also, the subjective nature of these indices is also viewed as a drawback by some economists.

Business confidence indicators are actively monitored by many economic agents, including not only managers and entrepreneurs, but also central banks, financial analysis organizations and traders. It is an indicator by whichimportant information about current and future economic conditions in various spheres of the economycan be disclosed by employees of the firms (Santero and Westerlund, 1996). Application of these indices is very beneficial as they can be calculated quite fast and it is believed to be more accurate than hard data (Holmes and Silverstone, 2010; Santero and Westerlund, 1996). They show in advance the signals of short-term changes in the economic activity, and moreover, the results of these indicators are subject only to minor revisions (Coetzee, 2014). In addition to that, sometimes it is more reliable to use business sentiment indicators' data than the official economic variables to understand the particular economic situation, because “the real-time picture of economic dynamics may differ in some sense from the same picture in its historical perspective, because all fluctuations receive their proper weights only in the context of the whole…” (Smirnov, 2010).So, these indices are sometimes being deemed as one of the best leading indicators of the real economy as the indices give useful predictions regarding the future fluctuations of the key economic variables (Dreger and Kholodilin, 2013). Confidence indicators are considered to be leading indicators because their decrease symbolizes a soon downturn in the economic activity and recession in the business cycle, whereas their increase represents economic growth and expansion of the business cycle (Dreger and Kholodilin, 2013). That is why confidence indices can be used to forecast economic growth and fluctuations in business cycles (Holmes and Silverstone, 2010; Santero and Westerlund, 1996).

In order to identify a coming downturn or a rise in economic activity surveys of company's managers are widely used. This method allows getting all the relevant economic information quite fast comparing to obtaining statistics from national accounts because for the latter it usually takes up to 4 months to gather, process and publish this data, whereas in case of managerial surveys, data is usually available in a month. For instance, in case of Russia, GDP(which is publishedby the Russian Statistical Agency-- Rosstat) is issued quarterly and a first estimate for the quarter GDP is released aroundthe fifteenth day of the second month of the next quarter.Information received from the surveys is considered a reliable source of detecting the changes in business environment and economy as a whole, as it depicts expectations and current state of affairs of the real economic agents. However, it is clearly a subjective information because it is based only on subjective perceptions of managers and not on the real figures. These surveys consist of questionnaires to managers of firms regarding actual and anticipated product prices, inputs prices, volume of output, different aspects of employment and investment activities, demand, suppliers, financial stability etc. In a typical survey, managers are responding to questions regarding firm and market conditions, so, they should highlight whether those conditions remain unchanged, have worsened or improved since the last month, and also give predictions about it for the next month. After that, the results for each question are being averaged to receive a general reflection of sentiments. There are typically more than 1000 firms being surveyed. Besides that, as a general rule, different types of firms are being represented there -- small, medium and large. Moreover, distinct surveys can be conducted for different sectors, reflecting the situation for each particular industry. Hence, by looking at the results of these questions, general trends in the economy can be seen, whether entrepreneurs experience and expect improving of today's situation or not. Each country has its own business confidence indices which reflect all the necessary features of this particular country's economy. There are a lot of organizations that conduct their own surveys and then construct different types of business activity indicators.

Implementation of a methodology of business confidence indices based on surveys is quite short for Russia, it is several times less than in the USA. System of business tendency surveys as a pilot project had been introduced in Russia for the first time in 1992 by the Centre for Economic Analysis, which was developed on the principles of the JointHarmonised EU Programme of Business and Consumer Surveys under the Tacis programme `Statistics 2, 3, 5' (Lipkind et al., 2018). But regular and consistent surveys and indicators' methodology have only seriously started to develop around the huge crisis in 1998, when the economy was in a massive prolonged recession, it became obvious that precise and credible confidence indicators are necessary for decision-making of the economic agents of the country, and what is more important, there was enough statistical data available to test the efficiency of these indicators empirically (Smirnov, 2019). Since that time Russia has begun to elaborate and try out different types of indicators in order to monitor business sentiments and economic activity, such as indicators based solely on tendency surveys of firms, and cyclical indicators, which are divided into leading, lagging and coincident ones (Smirnov and Kondrashov, 2018).

1.2 Methodology of the PMI

Purchasing managers' index (PMI) is one of the most popular and old business confidence indicators in the world. The PMI has been first established in the USA in 1982, by the U.S. Department of Commerce (DOC) paired with the Institute of Supply Management. Since then, its methodology has been adopted by many countries across the world and the relationship between this index and many economic variables has been extensively studied. The PMI for Russia is calculated by Markit Economics, who also computes this index for the other 25 countries.

There are two types of the PMI that are computed for two different economic sectors - manufacturing PMI and non-manufacturing or services PMI. Also there is a composite index, which indicates an overall sentiment in the economy. As this paper employs only manufacturing PMI in the empirical study, a further discussion will be focused on this type of PMI only. So, by definition, it is a monthly indicator based on survey data from manufacturing companies, this indicator is seasonally adjusted and is represented as a weighted composite diffusion index that consists of five components of economic activities in the manufacturing sector (Koenig, 2002; Pelбez, 2003b). The responses of the committee members mirror the economic activities of the manufacturing sector, nationally, ensuring a report that accurately represents current business conditions. Diffusion indexes have the properties of leading indicators and are convenient summary measures showing the prevailing direction of change and the scope of change (IHS Markit, 2014).

The methodology behind the PMI, which is described in the guide by the IHS Markit (2014), is the following: first of all, the data from surveys of companies' managers is collected at the 2nd half of the month. All the interviewees stay anonymous. In case of Russia, approximately 300 companies are surveyed for computation of the manufacturing PMI. The survey consists of questions about various aspects of business, namely: “about changes in production, new orders, new export orders, imports, employment, inventories, prices, price changes of commodities,lead times, and the timeliness of supplier deliveries in their companies comparing the current month to the previous month”. The participants should answer how all these parameters have changed since the last month. These parameters are gathered into five components that have the following weights in the diffusion index: 30% for new orders, 25% for production, 20% for employment, 15% for supplier deliveries, and 10% for inventories (Bretz, 1990). So, the participants should reply that either there is no change, or there is an improvement, or there is a worsening regarding these components of business. An important rule of the PMI methodology is that these responses could never be changed or adjusted after being collected. Then, percentages of positive, negative, and “no change” responses are calculated, and diffusion index is computed. Afterwards, the weighted diffusion index is seasonally adjusted.And, the final PMI results are published at the first business day of the month.

As to the manner in which the diffusion index is computed, it is calculated asasum of the percentage of positive answers and one-half of the percentage of “no change” answers:

where:

-- share of positive responses

share of "no change" responses

Therefore, the index value domain is from 0 to 100. For example, if all the respondents indicated a worsening in the business conditions, then the index would be 0. In the opposite case, if everyone reported an enhancement, then the index value would be 100. And if all the answers indicated “no change”, therefore the index would be 50. So, the general interpretation is that, if the PMI value is above 50, it means that there is an overall improvement in the economic situation, and when it is below 50, there is a weakening. Moreover, the closer is the index value to 50, the less is the change in the economic variables during that month. Whereas the exact value of 50 can have two meanings -- either all the surveyed managers responded “no change”, or the number of positive responses was matchingwith the number of negative ones.

The PMI is considered to be a leading indicator of the business cycle. So, its data can be viewed as a forecaster of future change in the business cycle. For instance, if there is a fall in the PMI after a prolonged period of growth, then it is a prediction of a downturn (recession) of the cycle. And in the opposite case, if the PMI shows growth after a period of decrease, then it is a signal of an expansion.

1.3 Methodology of the ECI

One of the firstly introduced indicators of business activity in Russia was index of business confidence by Rosstat. The Rosstat Entrepreneurial Confidence Index (ECI) is a Russian index of business activity, which has only formed a consistent methodology and has started to be published on a regular basis since 2005. It incorporates international practices of conduction of surveys and the subsequent construction of the business index. In particular, the Rosstat methodology is harmonized with the practices of European Commission.

This index is a business confidence indicator which shows aggregated status of business sentiment in different economic sectors, and it is based on managers' responses about present and expected production conditions. The main purpose of the ECI is to give an overall representation of current and anticipated (for 3 months ahead) business conditions in Russia and also to supply economic agents with premature information about the future fluctuations of the main economic variables.

Based on the information about this index by Rosstat (2019), this index is based on monthly managers' survey, where more than 3500 large and medium firms take part on an on-going basis (small firms are not included there, they are accounted on their own separate survey and there is a distinct business confidence index for them). The firms, which participate in the survey, are chosen according to the type of their economic activity, which is listed in the Russian Classification of Economic Activities (OKVED2). The survey takes place after the 10th day of each month and the results are published on 14th day of that month. All the individual responses have a particular weight which corresponds to the size of the respondent firm, this size is determined by the quantity of employees. This business indicator is an average sum of differences (also referred to as a "balances") of positive and negative answers about the expected production, actual demand and the current inventory of a finished product (that is taken with the opposite sign). So, diffusion index of the ECI looks as follows:

For expected production, balance is made by subtracting the share of responses that reported "decrease" from those who pointed out "increase". In case of demand and inventory, the balances are constructed by subtracting the portion of answers that say "lower than normal" from those who remark "higher than normal".

Values of the ECI index have an equilibrium at 0, it is a situation when there is an equal amount of positive and negative answers. This index can take negative values, in case when there is pessimistic attitude in the business community. And in the opposite scenario, the ECI takes positive values, when this attitude is optimistic.

An important advantage and distinctive feature of this indicator is that it is held for several economic industries, so the climate of Russian private sector could be studied in more details. They are the following: "Mining and Quarrying", "Manufacturing", "Organizations of Electricity, Gas, Steam and Air Conditioning Supply", "Construction" and "Retail Trade Organizations". In particular, the first three industries constitute the industrial production. The first surveys in all of the above-mentioned sectors have been carried out approximately since the end of the nineties, except the services sector, for which this survey had only been developed in 2012. Another advantage is a high response rate, which is above 90% at the average, because the participation is being obligatory and firms have to necessarily report primary statistical information in order to constitute the official statistical figures (Lipkind et al., 2018). It is written in the federal law "On Official Statistical Accounting and State StatisticsSystem in the Russian Federation" (Federal Law No. 282-FZ, 2007). So, non-response is a rare case and it only happens, for instance, if some of the respondent firms are liquidated (Lipkind et al., 2018).

A substantial drawback of the Rosstat ECI is that "the wholly comparable time-series of this index is not existent because of the several changes in classification and aggregation methods» (Smirnov, 2010). Due to that fact, there are 3 different parts of these time-series, where the last change has been made in 2009.

1.4 Methodology of the BCI

Another popular business activity index, which is applied in Russia, is the Business Confidence Indicator by The Organization for Economic Co-operation and Development (OECD). By the OECD (2019) definition: "The Business Confidence Indicator (BCI), sometimes also called Industrial Confidence Indicator, corresponds to the arithmetic average of seasonally adjusted net balances of production future tendency, order books levels and stocks of finished goods (with inverted sign) in the manufacturing sector". It can be represented by the following formula, where B stands for the word "balance":

This index is standardized, so that its figures among different countries could be compared. For Russia, the BCI is computed only for manufacturing sector, whereas for the OECD-members there are also indices for construction, services and retail trade. The survey data source for this index is claimed to be the E.T. Gaidar Institute for Economic Policy, and this data is comprised with the European Harmonization principles. But the indicator itself is calculated by the OECD. More than 1150 of manufacturing firms from all the Russian regions take part in the survey each month, where the response rate is being 65-70% (Korte, 2012).In the survey questionnaire respondents are supposed to report their current opinion and future expectations (within 3-4 month period) about different sides of business activity. Then, the BCI figures are published on the first business dates of the month.

In the survey respondents are asked about whether they experience and expect an increase, decrease or no change in such aspects as production tendency, orders, etc. However, in calculation of balances of the answers, "no change" responses are purposely omitted. So, the balances are computed by subtracting the share of "decrease" answers from the share of "increase" ones.

For the standardised BCI, value of 100 is a bar that represents the normal situation. Everything above this bar indicates a positive attitude towards current and business situation and expected improvements in the future. Whereas a value below 100 is a signal of dissatisfaction of current business conditions and anticipated future distortions in them.

1.5 Methodology of the RSBI

The Russia Small Business Index, or RSBI, is Russian newest business confidence index, that has been established since the 3rd quarter of 2014 by the "Opora Rossii" organization. This indicator is different from the other ones because it is made on the basis of tendency surveys of the "SMM" firms -- which is small, medium and micro firms. So, in general, purpose of that index, proposed by its publisher, is to measure and report sentiments and business expectations mainly for those SMM firms. Also, a distinguishable feature of the RSBI is that it is based on survey responses of firms from different economic sectors, which include trade, manufacturing and services (Opora Rossii, 2018).

Due to the information from Opora Rossii(2019), to construct this index, around 2000 companies from various economic sectors are surveyed at a quarter frequency, what corresponds to a 63% of total quantity of the SMM firms. Surveys are conducted at 24 regions of the country. Respondents are required to answer questions regarding 4aspects of business, namely: Sales, Employment, Access of Funding, Investments. So, composite index includes these aspects with the specific weights shown in the given formula:

Managers are asked to provide their opinion on these 4 dimensions of business, as well as expectations. However, the system of responses for this index is different from what other indices, considered in this paper, are using. There is not only simply negative, positive and "no change" answers, there is more gradation and details available. So, for instance, in case of Funding Access, users of the RSBI can see not just whether it is in general easy or hard to get a credit funding, but also what exactly share of firms does not need funding, or already has funding, or what portion of companies applied for a credit and received approval, or did not receive, etc.

The RSBI has an equilibrium value of 50 points, therefore, a higher value of this index would be interpreted as a sign of increase of business activity, and a value lower than 50 -- as a decline in it. Also, the publisher says, its values are seasonally adjusted, but the exact technique is not specified.

A substantial drawback of this indicator is that the publisher does not explain its methodology in details. For instance:

how exactly the balances of answers are calculated

whether there is a different weight for each size of the firm

at what period of time the data is being gathered

what time horizon is used for expectations of managers

It would be interesting to know the details of construction of this index to be able to compare methodology of different indicators. Also, it could reveal some possible weaknesses of this index.

2. Literature review

Effectiveness of business activity indicators are being studied and discussed by many theoretical as well as empirical papers. Many researches are devoted to investigation of how precise is the prediction of business activity changes by various indices, i.e. whether they are useful in determining of turning points in business cycles. Majority of the papers are devoted to the investigation of usefulness of the PMI, and much less studies are focused on the BCI, as the latter has been existed for a much shorter time period and it is calculated for a smaller number of countries. Talking about Russian indices, there is a significant scarcity of literature about business tendency indicators because they started to be incorporated just a few decades ago. And also, their efficiency mainly concerns only Russian economists, whereas in case of the PMI and BCI, which are implemented in many countries across the world, there is much higher number of economists and organizations that are interested in studying of these indicators.

There is a decent quantity of studies that examine the strength of the relationship between such indices and actual macroeconomic variables, so in other words, they investigate whether these indicators give accurate and valuable information about the economy. However, quality of the association between tendency indices and economic activity in a particular region can vary due to many factors such as time period of study, selected dating of turning points in the economic activity, different seasonal adjustment techniques that are applied to data set (Smirnov et al., 2017).

2.1 Relationship of the PMI with economic variables

A crucial part of the existing literature on business activity surveys consists of the studies about the Purchasing Managers Index, as it was invented a long time ago and it is implemented in a broad range of countries. There is much less literature devoted to the study of the PMI worldwide as this index has been adopted in other countries much later than in the USA. Therefore, the PMI is considered by many scientists a credible tool for forecasting of GDP and changes in business cycle.Below, a plenty of papers which examine the PMI from various perspectives and for different economies are reviewed.

Most of the research papers are devoted to the measuring of relationship of the PMI with other macroeconomic variables, especially, GDP and economic growth.

Many papers on that topic consider the US economy. The first study on the efficiency of the PMIand its relationship with macrovariables was performed by Torda (1985), a scientist from the U.S. Department of Commerce and Institute of Supply Management, who has originally developed this index and revealed that there is a close relationship between it and GDP. And in addition to that, the author has found that this indicator is capable of explaining around 60% of variation in annual GDP. Since then, high correlation between the PMI trend and GDP trend has been detected by many thorough examinations. For example, early classic papers such as Raedels (1990), Harris (1991),Niemira (1991), Klein and Moore (1998), Niemira and Zukowski (1998), Kauffman (1999) have all observed high association between the PMI and US GDP, and have determined that this business index can serve as a leading indicator for the US economic activity. Then, Koenig (2002) have found evidence that the American PMI forecasts the Industrial Production (IP) and GDP, and that there is a strong relationship present between the PMI and the federal funds rate, which is considered to be an important instrument of monetary policy of the Federal Reserve of the USA. A further investigation by Chien and Morris (2016) has proven a positive correlation between the PMI and economic growth in the US with a coefficient of 0.75, which coincides with the result obtained by Koenig (2002).In another study regarding the US, researchers have conducted an investigation on three different dimensions of confidence indicators (for the period from 1980 to 2005), namely, consumer, business (the PMI) and investor confidence indices, and their relationship with economic fluctuations measured by GDP, suggest that all of these sentiment indices are playing necessary roles in economic fluctuations (Afshar et al., 2011). The authors have made such a conclusion based on the fact that the hypothesis that these confidence measures do not Granger-cause GDP was rejected.

A recent study about South Africa, which was based on quarter time series data from 2000 to 2016, has identified a positive relationship between the PMI and economic growth and employment in manufacturing sector. The results were established based on the correlation analysis that has proved significant positive relationships between the above mentioned variables, and moreover, Bound's test for co-integration revealed a long-run relationship between the variables (Habanabakize and Meyer, 2017).

A study of manufacturing PMI for China revealed that it has a strong impact on important Chinese economic variables such as Producer Price Index, Consumer Price Index and the growth rate of secondary industry (He et al., 2015). Furtehrmore, Chien and Morris (2016) have shown evidence on strong positive relationship between the PMI and economic growth of China, they have found out a strong positive correlation with a coefficient of 0.73. The more recent study on expectation indicators and their relevance in Chinese economy have presented the evidence of the efficiency of the PMI (Olaleye, 2018). Seven different confidence indicators and their relationship to economic performance (proxied by GDP growth rate) were investigated for the period from 2007 to 2017, using correlation and Granger-causality analysis, OLS regression and Distributed Lag model. As a result, a positive relationship was highlighted between the PMI and the economic performance of China, therefore, the author suggested that this indicator can be used as a leading indicator and, at the same time, as a coincidental indicator.

A recent paper about Indian GDP has examined whether there is a strong relationship between the level data as well as growth rate data on overall GDP, manufacturing GDP, services GDP and the corresponding figures of overall, manufacturing and services PMIs (Das and Coondoo, 2018). For the years from 2006 to 2014, it was found that all three growth rates of GDP do not have a correlation with any of the PMIs, whereas for all the level GDP data, there is a significant positive correlation with the services PMI only. In addition to that, services PMI turned out to have the greatest marginal effect on manufacturing GDP.

A further research on the economy of Turkey has also pointed out that that there is a pretty huge association between the PMI and Turkish GDP, but what is more important, is that the threshold value of the PMI for this economy, which indicates a normal or average business situation, is not 50, but 47,5 (Eren, 2014). Therefore, a value of 50 (or anything above 47,5) for this economy would be a signal of the expansion of business cycle and optimistic sentiments - which is a crucial note for everyone who monitors this index.

2.2 Application of the PMI for forecasting

There is a substantial part of literature that examines the ability of the confidence indicators to predict business cycles in the economy, and especially, detecting recessions. Most of such papers prove that the confidence measures are indeed good for forecasting or, so called, nowcasting of the changes in economic activity. Basically, nowcasting is a prediction of the nearest future, usually it is a period within one quarter of the year. By a formal definition, a process of nowcasting is "a method of tracking the real-time flow of the type of information, i.e. it is a method of evaluating the marginal impact that intra-monthly data releases have on current quarter forecasts of macroeconomic variables" (Giannone et al., 2008). This method gives the possibility to detect the latent factors quite efficiently and, moreover, it allows to adjust the forecasts in real-time, as soon as new data on macroeconomic variables becomes publicly available (Giannone et al., 2008, 2010; Baсbura and Modugno, 2014). This approach is useful for "key macroeconomic variables which are collected at low frequency, typically on a quarterly basis, and released with a substantial lag; To obtain “early estimates” of such key economic indicators, nowcasters use the information from data which are related to the target variable but collected at higher frequency, typically monthly, and released in a more timely manner." (Baсbura et al., 2013). In most of the existing literature it is believed that implementation of the PMI gives pretty successful results for nowcasting of fluctuations of economic variables.

There is a substantial number of papers devoted to the estimation of nowcasting properties of the PMI in the global context. For example, one of such examinations has highlighted that this index is one of the best indicators for forecasting the worldwide economic activity (Drechsel et al.m 2014). The Global Composite PMI and manufacturing PMI were tested in this study along with the other leading indicators. As a result, the PMI for manufacturing turned out to be one of the two indicators that give substantial enhancements to forecasts of the International Monetary Fund.

Further, a study, devoted to the investigation of forecasting accuracy of the PMI regarding short-run real GDP growth in the United Kingdom, Eurozone, Japan and China, has also proven the significance of the PMIs (Godbout and Jacob, 2010). It turned out that the exploitation of such an indicator indeed improves the quality of in- and out-of-sample forecasts for the above-listed countries.

Further on, Chudik and Pesaran (2014) have conducted an investigation of 48 countries to determine the significance of the PMI for anticipating global growth, then forecast it applying the Global VAR (GVAR) method and, finally, compare the obtained forecasting performance with results of other methods. As a consequence, this work has showed that the PMIs are very valuable for nowcasting, irrespective of the implemented forecasting approach. However, it is noteworthy that the value added by these PMIs declines pretty fast as the forecast period increases.

Besides that, Rossiter (2010) has attempted to evaluate the significance of implementation of the global PMI for predicting quarter global output, inflation and import. The obtained results suggest that this index gives a plenty of useful information about the situation in the world economy and it is helpful for predictions of the aforementioned global macro-variables, because, in comparison to the performance of benchmark models, the latter give less accurate figures. However, when the author has tested the period from 2008 to 2009 (the time of so called "Great Recession"), he noticed that the PMI-augmented models were not able to reflect perfectly the rapid contraction in the worldwide economy.

Research on that topic in Eurozone countries also, mainly, leads to similar conclusions about effectiveness of the PMI. "Real-time evidence for the euro area shows that a tracker for real GDP growth using only the PMI composite output is of similar accuracy for the final GDP release as the first GDP release; No signs of instability--except during the 2008/09 crisis--in this tracking performance are found" (De Bondt, 2018). The other researchers, that studied this topic, applied mixed data-sampling (MIDAS) models in their study, where low-frequency variables (in this study, quarterly GDP growth) was matched to lags of high-frequency variables (Leboeuf and Morel, 2014). So, among more than 50 different indicators and variables for both Japan and Europe that were tested in MIDAS model, the PMI was considered the best in predicting real GDP growth in Europe, and the PMI along with the Economy Watchers Survey Indicator appeared to have the highest predictive power for the Japanese economy.

Another large research devoted to now-casting using the European PMI was conducted for all countries, which are in the membership of the Eurozone, covering 60 months from 1997 to 2011 (Lombardi and Maier, 2011). It was found that the PMI-based model as well as the dynamic factor model is more preferable to AR model. Also, one of the conclusions, that was drawn from the study, is that "as the dynamic factor model processes the information in this data set, it extracts a first factor that closely resembles the PMI, or put differently: the PMI turns out to be a good way to represent the data flow". However, the researchers claim that, on average, the factor model processes the information in a more efficient way than the PMI model.

Gajewski (2014) has studied that among the 4 largest economies of Europe, models with composite PMI give the most precise results only for such countries as Italy, France and Germany, as well as for the whole Euro Area as an aggregate, whereas for Spain it is not the case. The analysis was performed using cross-correlations and out-of-sample forecasting errors that have been processed by performing the rolling regressions in fixed-length window.

Similar results about the Eurozone were obtained by a more recent study, where efficiency of the PMI has been examined (Kilinc and Yucel, 2016). The forecasting abilities of the PMI on GDP growth in euro area for the period from 1998 to 2015 were studied by estimating a number of different specifications, which start with a pre-defined threshold. The results have showed that threshold of 50 is indeed a good bar that serves as a division between improvements and distortions in GDP. Moreover, the research has pointed out a conclusion that the PMI has a high forecasting power, because it can predict growth rate of GDP for the current and for the following quarters, also, this index can be used as a leading indicator.

Furthermore, another analysis on the strength of relationship between this index and the GDP growth for the 3 major European member states (namely, Italy, Germany and France) has revealed that the composite PMI has the highest correlation with GDP growth, compared to another popular sentiment indicator -- the European Commission's Economic Sentiment Indicator (ESI) (Smith, 2018). Also, from 2000 to 2018, the PMI has always showed an apparent superiority over the ESI in terms of correlation with GDP growth and nowcasting correctness, that were computed for the above mentioned 3 countries and the Eurozone as a whole. Nevertheless, in case of France, both of these indicators show the worst results. For example, the PMI has showed a correlation around 50% and the percentage of correct forecasting is 59% during the whole period.

Likewise, a recent analysis regarding Germany and Switzerland has highlighted that monthly indicators built on foreign PMI indices have a high correlation with quarter export growth of these two countries and, moreover, the comparison of forecasting models discerns that models with the incorporated PMI perform pretty successful compared to other benchmark models (Grossmann and Scheufele, 2016). This analysis was also held using MIDAS models, combined with performance of in-sample and out-of-sample tests.

Another study regarding Switzerland have also confirmed that there is a high correlation between the Swiss PMI and real GDP, and that the former has proper nowcasting abilities as almost all of the signs of fitted values match with the signs of the actual GDP growth rate figures (Mьller, 2013). Also, the nowcasting model with this index incorporated within has proved that it is able to detect turning points of the economic activity quite well, even during the 2008 crisis. However, the author says that this exact index is rather a coincident indicator, than a leading one.

In case of Italy, the efficiency of the PMI in forecasting was confirmed by (Aprigliano, 2011). She tested the association between the Italian PMI and the quarter growth rate of the Index of Industrial Production (IPI). The study has showed that this index can indeed capture the trends of the "medium-to-long-run" component in the IPI, what was the main question of the whole study.

A study regarding Germany has not revealed such a successful results of the PMI's performance in predicting the economic activity. Results of Granger-causality test and cross correlations have demonstrated that among 3 different business confidence indices applied in this country, the PMI has the lowest results in forecasting of industrial production (Hьfner and Schrцder, 2002).

Considering the studies of the US economy, a lot of research has been made. For instance, some researchers have proved the effectiveness of the PMI in forecasting movements of GDP in real time, as they calculated that this index “the current quarter forecast of GDP by about 12%, and the one-quarter-ahead forecast of GDP by 31%” (Harris et al., 2004). Tsuchiya (2012) used Fisher's exact test based on contingency tables to show that the forecasts of the PMI accurately predict the direction of change in GDP and Industrial Production.

Also, Lindsey and Pavur (2005) built a regression model that is based on “inherent cycles” of the PMI. They tested between 12 and 65 months of the PMI data to forecast turning points for the index and predict alterations in the business activity cycle. The authors say that "as a result, the model with the implemented PMI appeared to be superior to the more commonly used Box-Jenkins forecasting technique".

In another research paper, which tested forecasting effectiveness of two different indicators, the PMI and the Survey of Professional Forecasters(SPF), it was found by performing real-time out-of-sample test that both of these indicators are helpful in forecasting of GDP and Industrial Production growth (D'Agostino and Schnatz, 2012). But "the SPF clearly outperforms the PMI in forecasting GDP growth, while it performs quite poorly in anticipating Industrial Production growth".

Another study about the American economy has shown that the PMI along with other indices computed by the Institute for Supply Management (ISM) has a big role in helping precisely calculate the nowcast ("which is the task of predicting the present, the very recent past, or the very near future of GDP") of current quarter GDP in the USA, even though there are a lot of other indices and variables existing (Lahiri and Monokroussos, 2013). The authors suggest that such a great usefulness of the PMI and other ISM indicators is based on the fact that these measures are available in the month ahead of other existing variables. Also these economists proved that marginal contribution of the PMI in anticipating GDP growth is quite high.

Furthermore, large study by Christiansen and Eriksen (2014) has investigated the forecasting features of the PMI, comparing it with more than 150 classical financial and macroeconomic predictors measured monthly and the data covered the period from 1978 to 2011. The researchers tested the index by using common factor approach, that have shown that the PMI is strongly statistically significant and it is the best recession predictor out of all the variables tested for both in-sample and out-sample models. Important conclusion drawn from this study is that “sentiment-based variables contain considerable higher predictive power than the classical recession predictors". And what is also noteworthy, "combining the sentiment variables with the classical recession predictors provide strong predictive power for future recession periods”. However, these authors say that they are not able to justify fully why sentiment indicators are so good at forecasting recession.

Another study devoted especially to the 2008-2009 recession have also revealed that the PMI (along with 2 other different cyclical indicators) can be considered a credible and rather informative predictor of turning points in economic activity in real time, because this cyclical indicator has altered its trajectory in the opposite direction some months prior to a turning point (Smirnov, 2011). In particular, it gave a drastic sign of the forthcoming recession of 2008 in December of 2007 and also it showed a clear signal of its end in July of 2009. Among the results of other 2 studied cyclical indicators, the PMI have showed the most precise real time predictions.

Analysis of relationship of both Canadian and the US confidence indicators on forecasting of GDP of Canada has revealed that the exploitation of the US data also significantly improves performance of such prediction models, and moreover, the best performance is achieved when data from both of these indicators is employedsimultaneously (Moran et al., 2018).

In case of China, which has only implemented the PMI methodology in 2005, this index has always been considered as a successful and reliable predictor of future economic activity. For instance, Zhang and Feng (2012) have proved that this confidence indicator is "ahead of GDP trend for almost a quarter, and hence, is fully capable to predict the economic trends and turning points".

An analysis of the PMI carried out for India from 2005 to 2012 have demonstrated that this index is a reliable tool for forecasting changes in the Indian Wholesale Price Index as it has significant predictive power (Khundrakpam and George, 2012). The authors applied OLS regression estimates and the ARDL method in their study. Another study on that topic have shown that "PMI based model is found to perform better than the autoregressive model and some other models in a data rich environment, in anticipating near-term GDP growth" (Bose, 2015).

Examination of forecasting process for Turkish economy have determined that nowcasting model, that contains the PMI data, predicts GDP growth much better than models that simply include Industrial Production, import and export quantity indices (Akkoyun and Gunay, 2012). In this study out-of-sample forecasting exercise was conducted for the quarter data from 2008 to 2012, combined with application of the Stock and Watson coincident indicator.

2.3 Critique of the PMI

Although many researchers believe that there are a lot of advantages in the concept of PMI, there are some people who think this methodology is not so perfect and credible. One of the biggest critiques is that this index does not take account for the influence that a few very large and substantial firms could possibly have on the whole market or even the whole economy (Porter, 2002). This situation happens because the methodology of the PMI implies that all firms are considered equal in terms of size. Also, Porter (2002) pointed out that another drawback of this index is that it only gives the conclusion for the whole country, there is no particular figures being calculated for each of the country's regions. It is a problem as each region can have its own unique economic situation.