- Our estimate of GDP won’t regain pre-coronavirus level till 2023
- On official readings, the economy is set to recover by start of 2022
- We see 1% qoq growth over next year vs 1.7% using official data
- New 4trn yuan stimulus package focuses on cutting business costs
- But bond issuance also booming to fund investment spree
- Spending to revive growth will be much less effective than after GFC
- Track Enodo’s new nowcasting model to see how the economy responds
Analysing China’s economy has always been fraught with difficulty. But Covid-19 has made the job that much harder, not least as a key dataset we used to produce the Enodo real GDP growth estimate was discontinued earlier this year. But that just makes us even more determined to figure out where the economy is headed at this critical juncture.
Data, data, data… we collect as many pieces of the puzzle as possible and try to build a coherent picture of what’s going on in the economy. We resolved the latest problem by finding a suitable price series to substitute for the now discontinued fixed asset investment price data we used to deflate nominal investment spending.
But, more importantly, in this note we introduce to you Enodo’s nowcasting model for forecasting China’s GDP.
Its main message is a sobering one for Beijing and could well extend to the rest of the world. On current readings, our model suggests that it will take until the start of 2022 for real output to regain its pre-Covid-19 level.
That is going by the official data. But going by the way Enodo constructs its estimates, Beijing will have to wait till the beginning of 2023 for real GDP to get back to where it was before the pandemic struck.
Beijing just unveiled a comprehensive stimulus at the delayed National People’s Congress. There’s more to the plan than meets the eye. On the face of it, the authorities are trying to show that they’re making a strong and responsible effort to lift the economy out of its coronavirus-induced slump.
Beijing has pledged a record Rmb4trn in cost cuts, including tax exemptions, lower bank rates, waived contributions to social welfare funds and reduced prices for utilities such as electricity. This contrasts with the Rmb4trn stimulus package in response to the Global Financial Crisis, which was focused on debt-fuelled state spending.
But Premier Li Keqiang also announced Rmb1trn in additional spending to be raised from a 0.8 percentage point increase in the central government’s fiscal deficit ratio, taking it to 3.6% of GDP, along with Rmb1trn from sales of special treasury bonds.
In addition, Beijing has also given local governments approval to issue another Rmb1.6trn of special purpose bonds that will be used to fund infrastructure construction. Behind the scenes, the authorities have also been putting pressure on state-owned enterprises to ramp up investment.
How effective China’s efforts will turn out to be, and how they will alter the trajectory the data are indicating, remains to be seen. We have argued that every yuan thrown at the economy now will produce that much less bang than post-2008. This is where the Enodo nowcasting model comes in, allowing us to monitor a large set of key data and what they tell us about GDP growth in real time.
The nowcasting model extracts the latent factors driving movements in the data and produces a forecast of each economic series that it tracks: when the actual release for that series differs from the model's forecast, this`news' impacts the nowcast of GDP growth.
According to the model, based on how Enodo estimates GDP we can expect no real growth in Q2 followed by a gradual recovery over the next four quarters to about an average of 1% qoq. On the official GDP data, the rebound in H2 this year and H1 next year is stronger, averaging 1.7% qoq.
You can read more about how we construct the model in the Appendix below. We will soon be adding it to our website for you to monitor in real time as the various series that make up the model are updated each month.
Appendix
Enodo Nowcasting Methodology
The nowcast methodology we use is based on the (parametric) factor model developed by Giannone, Reichlin and Small (2008). Their estimation procedure exploits the fact that relevant data series, although they may be numerous, co-move quite strongly. That means their dynamics can be captured by few common factors.
In other words, all the variables in the information set are assumed to be generated by a dynamic factor model (DFM), which copes effectively with the so-called ’curse of dimensionality’ (large number of parameters relative to the sample size).
The empirical model can be summarised in the following equation:
In order to nowcast Chinese GDP, we have collected a panel of 48 macroeconomic and financial series. Most of the data is seasonally adjusted by Enodo. For a full list of variables see Table 1 below. The frequency of the series is either monthly or quarterly and is denoted in the second column.
Each variable enters the model in stationary form: in most cases this requires simply including the series in the same units as it is published and tracked by markets. For other variables we calculate first differences or percentage changes. These transformations are appropriate for the purpose of producing nowcasts for the quarterly growth rate of China’s GDP.
According to the chosen specification, in the model there is a global factor (G) which affects all variables. In addition, a few local blocks are included to control for idiosyncrasies in particular subgroups of series; this can improve inference even though the model is robust to the presence of local correlations.
Specifically, to model the local correlations in survey data, we include the soft block (S), on which only variables representing economic agents' perceptions and sentiments load. Four additional local blocks are included for real (R), exchange rate (E), monetary (M) and price (P) variables. The information about factor loadings can be found in column ‘Block’.
Series Name | Frequency | Block | Category | |||||
G | S | P | R | E | M | |||
Electricity Production | m | . | 0 | 0 | . | 0 | 0 | Real |
Steel production | m | . | 0 | 0 | . | 0 | 0 | Real |
GDP | q | . | 0 | 0 | . | 0 | 0 | Real |
GDP deflator | q | . | 0 | . | 0 | 0 | 0 | Prices |
CPI | m | . | 0 | . | 0 | 0 | 0 | Prices |
CPI (non food) | m | . | 0 | . | 0 | 0 | 0 | Prices |
CPI (food) | m | . | 0 | . | 0 | 0 | 0 | Prices |
PPI | m | . | 0 | . | 0 | 0 | 0 | Prices |
Power General | m | . | 0 | 0 | . | 0 | 0 | Real |
Value Added of Industry | m | . | 0 | 0 | . | 0 | 0 | Real |
Total FAI Volume | q | . | 0 | 0 | . | 0 | 0 | Real |
Total Construction | q | . | 0 | 0 | . | 0 | 0 | Real |
Residential Buildings Construction | q | . | 0 | 0 | . | 0 | 0 | Real |
Office Buildings Construction | q | . | 0 | 0 | . | 0 | 0 | Real |
Commercial Buildings Construction | q | . | 0 | 0 | . | 0 | 0 | Real |
Other Construction | q | . | 0 | 0 | . | 0 | 0 | Real |
Exports Ordinary Trade | m | . | 0 | 0 | . | 0 | 0 | Real |
Exports Processing & assembly | m | . | 0 | 0 | . | 0 | 0 | Real |
Exports Processing with imported materials | m | . | 0 | 0 | . | 0 | 0 | Real |
Exports Outward processing | m | . | 0 | 0 | . | 0 | 0 | Real |
Imports Ordinary Trade | m | . | 0 | 0 | . | 0 | 0 | Real |
Imports Processing & assembly | m | . | 0 | 0 | . | 0 | 0 | Real |
Imports Processing with imported materials | m | . | 0 | 0 | . | 0 | 0 | Real |
Imports Outward processing | m | . | 0 | 0 | . | 0 | 0 | Real |
Import Equipment imported for processing | m | . | 0 | 0 | . | 0 | 0 | Real |
Imports Equipment imported for processing assembly | m | . | 0 | 0 | . | 0 | 0 | Real |
Exports | m | . | 0 | 0 | . | 0 | 0 | Real |
Imports | m | . | 0 | 0 | . | 0 | 0 | Real |
NEER | m | . | 0 | 0 | 0 | . | 0 | Exchange Rate |
REER | m | . | 0 | 0 | 0 | . | 0 | Exchange Rate |
USD/CNY | m | . | 0 | 0 | 0 | . | 0 | Exchange Rate |
Interbank Offered Rate-Overall | m | . | 0 | 0 | 0 | 0 | . | Monetary |
Interbank Offered Rate-Overnight | m | . | 0 | 0 | 0 | 0 | . | Monetary |
Interbank Offered Rate-7 day | m | . | 0 | 0 | 0 | 0 | . | Monetary |
1-year real household saving rate CPI | m | . | 0 | 0 | 0 | 0 | . | Monetary |
1-year real lending rate PPI | m | . | 0 | 0 | 0 | 0 | . | Monetary |
CN: Purchasing Managers' Index: Mfg | m | . | . | 0 | 0 | 0 | 0 | Soft |
CN: PMI: Mfg: Production | m | . | . | 0 | 0 | 0 | 0 | Soft |
CN: PMI: Mfg: New Order | m | . | . | 0 | 0 | 0 | 0 | Soft |
CN: PMI: Mfg: New Export Order | m | . | . | 0 | 0 | 0 | 0 | Soft |
CN: PMI: Mfg: Backlog Order | m | . | . | 0 | 0 | 0 | 0 | Soft |
CN: PMI: Mfg: Finished Goods Inventory | m | . | . | 0 | 0 | 0 | 0 | Soft |
CN: PMI: Mfg: Purchases | m | . | . | 0 | 0 | 0 | 0 | Soft |
CN: PMI: Mfg: Import | m | . | . | 0 | 0 | 0 | 0 | Soft |
CN: PMI: Mfg: Purchasing Price Index | m | . | . | 0 | 0 | 0 | 0 | Soft |
Consumer Confidence Index | m | . | . | 0 | 0 | 0 | 0 | Soft |
Consumer Satisfactory Index | m | . | . | 0 | 0 | 0 | 0 | Soft |
Consumer Expectation Index | m | . | . | 0 | 0 | 0 | 0 | Soft |
Retail sales volume | m | . | 0 | 0 | . | 0 | 0 | Real |