<abstract xmlns="http://www.w3.org/1999/xhtml"><p>Macroeconomic forecasters are often believed to idealistically work on improving the accuracy of their estimates based on for example the Root Mean Squared Error (RMSE). Unfortunately, reality is far more complex. Forecasters are not awarded equally for each of their estimates. They have their targets of acquiring publicity or to earn prestige. This article aims to study the results of Parkiet's competitions of macroeconomic forecasting during 2015–2019. Based on a logit model, we analyse whether more accurate forecasting of some selected macroeconomic variables (e.g. inflation) increases the chances of winning the competition by a greater degree comparing to the others. Our research shows that among macroeconomic variables three groups have a significant impact on the final score: inflation (CPI and core inflation), the labour market (employment in the enterprise sector and unemployment rate) and financial market indicators (EUR/PLN and 10-year government bond yields). Each group is characterised by a low disagreement between forecasters. In the case of inflation, we found evidence that some forecasters put a greater effort to score the top place. There is no evidence that forecasters are trying to somehow exploit the contest.</p></abstract>

Macroeconomic forecasters are often believed to idealistically work on improving the accuracy of their estimates based on for example the Root Mean Squared Error (RMSE). Unfortunately, reality is far more complex. Forecasters are not awarded equally for each of their estimates. They have their targets of acquiring publicity or to earn prestige. This article aims to study the results of Parkiet's competitions of macroeconomic forecasting during 2015–2019. Based on a logit model, we analyse whether more accurate forecasting of some selected macroeconomic variables (e.g. inflation) increases the chances of winning the competition by a greater degree comparing to the others. Our research shows that among macroeconomic variables three groups have a significant impact on the final score: inflation (CPI and core inflation), the labour market (employment in the enterprise sector and unemployment rate) and financial market indicators (EUR/PLN and 10-year government bond yields). Each group is characterised by a low disagreement between forecasters. In the case of inflation, we found evidence that some forecasters put a greater effort to score the top place. There is no evidence that forecasters are trying to somehow exploit the contest.

Macroeconomic forecasting in Poland: The role of forecasting competitions

SGH Warsaw School of Economics, Collegium of Economic Analysis

Central European Economic Journal

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

{"article-title":"Macroeconomic forecasting in Poland: The role of forecasting competitions"}

Macroeconomic forecasters are often believed to idealistically work on improving the accuracy of their estimates based on for example the Root Mean Squared...

Goodness-of-Fit test statistics

	success_i,t = 0		success_i,t = 1

Percentile	Actual	Expected	Actual	Expected	H-L Value
1	12	11.99	0	0.01	0.01

2	12	11.99	0	0.01	0.01

3	12	11.90	0	0.10	0.10

4	12	11.80	0	0.20	0.20

5	11	11.49	1	0.51	0.50

6	11	11.29	1	0.71	0.12

8	8	8.24	4	3.76	0.02

9	6	6.05	6	5.95	0.00

10	1	1.27	12	11.73	0.06

Total	96	96.18	25	24.82	1.47

H-L Statistic		1.47	Prob. Chi-Sq (8)		0.99

Andrews Statistic		55.82	Prob. Chi-Sq (10)		0.00

Year	PMI index	Industrial production	CPI	Unemployment rate
2019	3.65	4.71	4.82	7.76

2018	3.13	5.65	4.95	7.15

2017	3.79	5.54	4.45	5.85

2016	2.90	4.72	3.79	7.34
2015	3.10	4.91	3.16	6.28

Year	PMI index	Industrial production	CPI	Unemployment rate
2019	1.44	1.68	1.70	3.02

2018	0.73	1.18	1.40	1.52

2017	0.80	1.19	1.19	1.14
2016	0.89	1.22	1.18	1.99

Model parameters – Eq. (5)

Variable	Coefficient	Std. Error	z-Statistic	Prob.
CPI	0.94	0.44	2.15	0.03

Core inflation	1.25	0.45	2.75	0.01

Unemployment rate	1.18	0.45	2.63	0.01

Employment	1.53	0.48	3.21	0.00

CA	1.67	0.47	3.58	0.00

EUR/PLN	0.91	0.53	1.71	0.09

POLGB10	1.46	0.41	3.58	0.00

Constant	−3.18	0.54	−5.87	0.00

Model diagnostics

McFadden R-squared		0.51	Obs with Dep = 0	96

LR statistic		62.57	Obs with Dep = 1	25
Prob (LR statistic)		0.00	Total obs	121

Variable	β₁	β₀	Prob (x1 = 0)	Prob (x1 = 1)
CPI	0.70	−1.02	15%	38%
	0.37 (0.06)	0.18 (0.00)

Core inflation	0.83	−1.09	17%	38%
	0.38 (0.03)	0.18 (0.00)

CA	0.66	−0.94	17%	39%
	0.35 (0.06)	0.18 (0.00)

Export	0.23	−1.00
	0.38 (0.54)	0.18 (0.00)

Import	0.64	−0.96
	0.36 (0.08)	0.18 (0.00)

Construction output	0.19	−0.78
	0.36 (0.59)	0.17 (0.00)

Production	0.07	−0.91
	0.41 (0.86)	0.17 (0.00)

Retail sales	0.07	−0.83
	0.37 (0.86)	0.17 (0.00)

Employment	−0.74	−0.59
	0.41 (0.07)	0.17 (0.00)

Unemployment rate	0.02	−0.78
	0.37 (0.96)	0.17 (0.00)

M3	0.65	−0.99	16%	37%
	0.35 (0.06)	0.18 (0.00)

PMI	0.35	−0.89
	0.36 (0.33)	0.17 (0.00)

EUR/PLN	−0.44	−0.74
	0.43 (0.3)	0.17 (0.00)

USD/PLN	0.02	−0.74
	0.37 (0.95)	0.17 (0.00)
POLGBs 10Y	−0.13	−0.71
	0.36 (0.72)	0.17 (0.00)

Macroeconomic forecasting in Poland: The role of forecasting competitions

Published Online: Sep 10, 2020

Page range: 1 - 11

DOI: https://doi.org/10.2478/ceej-2020-0001

Keywords
forecasting, strategic behaviour, incentives, Parkiet

© 2020 Jakub Rybacki, published by Sciendo

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Fig. 1

Fig. 2

Median score in selected categories of Parkiet's competition

Standard deviation of scores in selected categories of Parkiet's competition

Which variables increase the odds of scoring top five in Parkiet's competition

Are some analysts persistently better than others in forecasting? Models’ estimations output of Eq. (6)

Macroeconomic forecasting in Poland: The role of forecasting competitions

Published Online: Sep 10, 2020

Page range: 1 - 11

DOI: https://doi.org/10.2478/ceej-2020-0001

Keywordsforecasting, strategic behaviour, incentives, Parkiet

© 2020 Jakub Rybacki, published by Sciendo

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Fig. 1

Fig. 2

Median score in selected categories of Parkiet's competition

Standard deviation of scores in selected categories of Parkiet's competition

Which variables increase the odds of scoring top five in Parkiet's competition

Are some analysts persistently better than others in forecasting? Models’ estimations output of Eq. (6)

Keywords
forecasting, strategic behaviour, incentives, Parkiet