Firms Learn from Their Supply Chain to Forecast Inflation

Keyword: 
Inflation
Topic: 
Macroeconomics - Economic growth - Monetary Policy

Leveraging a unique dataset on Chilean firms merging expectation surveys and records from the VAT and customs registries, this column documents that firms’ aggregate inflation expectations depend on the prices at which firms purchase inputs from their suppliers. These findings reject the full-information rational-expectations hypothesis and are consistent with facts associated with information rigidity. Estimates of a firm-level Phillips’ curve suggest that firms' aggregate inflation expectations fully pass-through to sales prices. We argue that this channel has the potential to lead to dispersion of inflation expectations, price dispersion, and weaken the expectation channel of policies.

The pandemic, together with policy interventions aimed at boosting demand, generated supply chain bottlenecks that put upward pressure on prices in many countries (WEO October, 2021; WEO April, 2022). High levels of inflation, in turn, prompted many central banks around the world to engage in a tightening cycle to ensure that inflation expectations remain anchored to the target. But how do firms—the actual price setters in the economy—form inflation expectations in this context? In our work (Albagli et al., 2022), we use a unique dataset merging expectation surveys of Chilean firms with administrative records from the VAT and customs registries to shed light on (i) how prices observed along the supply chain and firms’ inflation expectations are related, and (ii) if inflation expectations affect firms’ pricing decisions.

Disagreement, inattention, and the supply chain

Expectation surveys reveal that Chilean firms have substantially different views about next year's CPI inflation, consistent with the evidence on advanced economies (Coibion et al., 2018). Cross-sectional dispersion of firms' expectations is generally wide, even though it varies over time, narrowing when actual CPI inflation converges towards the central bank's target and widening when it deviates from it (Figure1, panel (a)). Despite such disagreement across forecasters, on average firms' predictions tend to correlate well with the inflation outcome.

Another key fact is the high degree of inattention to macroeconomic developments (Mackowiak et al., 2009; Pasten et al., 2016). In principle one would expect firms to use economy-wide information to form their beliefs about aggregate inflation, while idiosyncratic shocks should be ignored. As a result, disagreement about future inflation would tend to zero and inattention to macroeconomic news would be low. However, we find that in response to a change in CPI inflation half of the firms do not alter their inflation predictions (Figure1, panel (b)), and one fifth of them predicts a change in inflation in the opposite direction, suggesting that other factors influence the way in which firms form their expectations.

To explain these facts, the literature points to information frictions which prevent firms from accessing and adequately processing the data (Mankiw and Reis, 2002; Sims, 2003). In our paper we posit that such frictions lead firms to use changes in prices at which they source inputs from the suppliers to form their expectations about aggregate inflation. That is, as in Lucas (1972) firms operate as if they were located on different islands and learn from a subset of islands they trade with. If this is true, forecast disagreement may then arise because firms would rely on ‘local’ conditions, which are not necessarily the same across firms and that may not even have an aggregate effect. This, in turn, would make firms inattentive to inflation developments because they would deem aggregate information less relevant than supply chain information for their business.

Notes: In panel (a) the blue line denotes the median of firms' expectations about CPI inflation, the shaded areas denote the cross-firm interquartile range (dark blue) and the cross-firm interdecile range (light blue), and the red line denotes CPI inflation. In panel (b), the red and blue bars denote the shares of firm-month observations that report a decline or an increase in inflation expectations a month after a change in CPI inflation, where a change in CPI inflation is defined as a variation larger than half of its standard deviation; the gray bars denote the share of firm-month observations that report unchanged inflation expectations.

Supply chain inflation and aggregate inflation expectations

To test if firms use input prices to form their beliefs about future inflation, we construct a measure that tracks how expensive are the inputs that firms buy from their suppliers, which is effectively a metric of inflation along their supply chain. As firms buy different products from different suppliers, purchase prices display substantial dispersion across firms and we find them to be highly volatile over the sample period. (Figure 2, panel (a)). Yet, despite cross-firm heterogeneity and time volatility, we find that input price inflation and firms’ expectations of aggregate inflation are correlated (Figure 2, panel (b)).

Notes: Panel (a) presents the cross-firm distribution of input price inflation, where the blue line denotes the median and the shaded areas denote the cross-firm interquartile range (dark blue) and the cross-firm interdecile range (light blue); the red line denotes actual CPI inflation. Panel (b) presents a binned scatter plot in which each dot represents the average of input price inflation and firms’ inflation expectation, and the line denotes the linear fit.

We then examine if changes in purchase prices influence firms’ expectations about aggregate inflation. We narrow our focus to changes in purchase prices that are unrelated to the economy’s inflation, which allows to directly test if firms form expectations rationally. If firms were rational, they would discard the information coming from shocks in supply chain inflation that do not have aggregate effects when forecasting aggregate inflation. However, in presence of information frictions, firms may not be able to distinguish between shocks that have aggregate effects and those that do not, or they may only care about shocks that have immediate consequences for their businesses. As a result, firms would end up extrapolating some aggregate signal from unrelated changes in supply chain prices when forecasting aggregate inflation.

We estimate that a one standard deviation increase in supply chain inflation leads firms to revise upward their aggregate inflation expectations by 0.1 percentage point (Figure 3, panel (a)), which is about one fourth of the magnitude associated to a same-size-shock in CPI inflation (Figure 3, panel (b)). These results are remarkable given that our specification uses variation in supply chain inflation that is orthogonal to CPI inflation outcomes. As for the timing of the responses, the effect of an increase in supply chain inflation dies out over a 14-month period, likely reflecting the time needed for firms to realize that these changes in input prices are not determining aggregate inflation. On the other hand, the response to an increase in CPI inflation is shorter-lived, possibly owing to the credibility that the Chilean central bank gained over the past few decades. These results add to the literature rejecting the full-information rational-expectation hypothesis (Coibion et al., 2012; Coibion et al., 2015; Bordalo et al., 2020). [1]

Price-setting behavior of firms

After establishing that firms form their expectations about aggregate inflation based on the prices at which they buy inputs from their suppliers, we show that in fact they set their prices according to these beliefs. Departing from the vast literature estimating the Phillips curve with aggregate or regional data, we estimate it at the firm level. To the best of our knowledge, estimations of the Phillips curve at the firm level are virtually non-existent in the literature.

Relying on direct measures for firms’ sales price inflation, inflation expectations, and marginal costs, we find that firms’ inflation expectations are a critical factor shaping their pricing decisions. In particular, we document a complete pass-through of changes of inflation expectations to firms’ sales prices. Under different specifications, including allowing for sluggish firms’ price dynamics, we show that firms’ price increases depend on expected future price increases with a coefficient which is not statistically different from one, pointing to a price-setting behavior as predicted by forward-looking New Keynesian models.

Implications

These findings offers a number of policy insights. If firms form their inflation beliefs according to price changes observed along the supply chain and these are heterogeneous across firms, the channel we highlight in this paper can lead to dispersion of inflation expectations. Also, as firms act on the basis of their beliefs by setting prices for their goods and services accordingly, forecast disagreement can translate into welfare-costly price dispersion. Finally, our results imply a weaker effectiveness of the expectation channel of policies. We argue that improvements in central bank communication aimed at reducing firms' inattention have the potential to dampen the effects of the information frictions highlighted in this paper. In this regard, experimental studies examining the effects of the type, amount, and the way in which information is communicated can be informative.

Notes: The horizontal axes denote the number of months after the shock, the lines denote the point estimates, and the shaded areas correspond to 90 percent confidence intervals computed with standard errors clustered at the firm and time level.


References:

Albagli, Elias, Francesco Grigoli, and Emiliano Luttini. 2022. “Inflation Expectations and the Supply Chain, IMF Working Paper 22/161, International Monetary Fund.

Andrade, Philippe, Olivier Coibion, Erwan Gautier, and Yuriy Gorodnichenko. 2022.“No firm is an island? How industry conditions shape firms’ expectations.” Journal of Monetary Economics, 125: 40–56.

Bordalo, Pedro, Nicola Gennaioli, Yueran Ma, and Andrei Shleifer. 2020. “Overreaction in macroeconomic expectations.” American Economic Review, 110(9): 2748–82.

Coibion, Olivier, and Yuriy Gorodnichenko. 2012. “What can survey forecasts tell us about information rigidities?” Journal of Political Economy, 120(1): 116–159.

Coibion, Olivier, and Yuriy Gorodnichenko. 2015. “Information rigidity and the expectations formation process: A simple framework and new facts.” American Economic Review, 105(8): 2644–78.

Coibion, Olivier, Yuriy Gorodnichenko, and Saten Kumar. 2018. “How do firms form their expectations? New survey evidence.” American Economic Review, 108(9): 2671–2713

Lucas, Jr Robert. 1972. “Expectations and the Neutrality of Money.” Journal of Economic Theory, 4(2): 103–124.

Mackowiak, Bartosz, and Mirko Wiederholt. 2009. “Optimal sticky prices under rational inattention.” American Economic Review, 99(3): 769–803.

Mankiw, N Gregory, and Ricardo Reis. 2002. “Sticky Information versus Sticky Prices: A Proposal to Replace the New Keynesian Phillips Curve.” Quarterly Journal of Economics, 117(4): 1295–1328.

Pasten, Ernesto, and Raphael Schoenle. 2016. “Rational inattention, multi-product firms and the neutrality of money.” Journal of Monetary Economics, 80: 1–16.

Sims, Christopher A. 2003. “Implications of Rational Inattention.” Journal of Monetary Economics, 50(3): 665–690.

World Economic Outlook. 2022. “War Sets Back the Global Recovery.” Chapter 1, April.

World Economic Outlook. 2021. “Recovery During a Pandemic.” Chapter 1, October.


[1] Previous related research analyzed the effect of industry inflation on aggregate inflation expectations, finding positive effects (Andrade et al., 2022). Compared to that, the data at hand brings about some key advantages. First, we interpret our results causally, as input prices are generally exogenously determined with respect to firms' expectations of CPI inflation. Second, our measure of supply chain inflation aligns well with Lucas’s notion of ‘islands’ as it directly quantifies price changes that firms observe in their transactions. Third, the quantitative nature of the expectation question allows us to obtain an estimate for the impact of a shock in supply chain inflation on firms' expectations for CPI inflation. Fourth, inflation expectations are measured over a 1-year horizon (rather than 3 months), which is closer to the one relevant to monetary policy. And fifth, the higher frequency of the survey mitigates the concerns that confounding factors may be biasing the estimates.

 

Share this