Research

This paper examines which macroeconomic signals shape household expectations and finds that unemployment shocks affect households expectations at least as much as inflation shocks. Using daily data on expectations, we identify announcement driven revisions and construct shock series under two benchmarks: sophisticated and naive. We demonstrate that labor market information significantly influences not only households’ subjective expectations about the economy, but also their expectations about inflation. Even in periods when inflation is decreasing, rising unemployment can significantly influence expectations. We also find evidence of state dependence, with households responding more to the variable that has been worsening.

This paper investigates the prevalence and characteristics of poor hand-to-mouth (P-HtM) and wealthy hand-to-mouth (W-HtM) households in India — an emerging economy where high household savings coexist with limited liquidity and weak credit access. Using a harmonized dataset that combines two nationally representative surveys, we construct household-level balance sheets including income, consumption, and liquid and illiquid assets. We classify households into poor and wealthy HtM categories following the methodology of Kaplan et al. (2014), and use a range of machine learning models to impute missing income data. Our findings indicate that while P-HtM households account for 2–5% of the population, W-HtM households comprise a much larger share — 15–27%, with total HtM prevalence ranging from 17–32%. Classifications based solely on net worth substantially understate this share. We also find that average propensities to consume are similar across HtM and non-HtM groups, highlighting the broader financial constraints facing Indian households. These results underscore the importance of distinguishing between liquidity and net worth when evaluating consumption behavior and the transmission of fiscal and monetary policy in developing economies.

This paper provides new empirical evidence on the effects of monetary policy shocks on household consumption, income, and employment in a large developing economy. Using high-frequency identification of monetary surprises combined with local projection methods, we estimate dynamic impulse responses to both current and expected policy shocks. Our results indicate that a contractionary shock to the short-term policy rate raises consumption and income on impact but reduces them in the medium run, while employment declines persistently. In contrast, a contractionary shock to the expected path of future interest rates increases consumption and employment but lowers income. We also observe heterogeneity across socio-economic groups: rural households, those with lower education, women, younger and older workers, and lower-caste groups exhibit significantly larger consumption declines. A back-of-the-envelope calculation yields a marginal propensity to consume of about 40 percent out of transitory, policy-induced income changes. Our findings highlight the importance of distributional channels in shaping the aggregate transmission of monetary policy in developing economies.

The savings rate in the Indian economy has declined sharply since 2007 and approached a low of 29% of GDP in 2020. In this paper, we take a closer look at the historical evolution of aggregate savings in India, as well as its different components. We use a simple framework that seeks to explain the time variation in savings over the sample period. We find that increased prosperity more than explains the decline in savings of Indian households, while the decline in old age dependency was a major offset. Our baseline forecast implies a roughly 2 percentage point decline in household savings as a fraction of GDP over the next five years.

India has completed a decade of flexible inflation targeting, and headline inflation has spent most of that decade inside the 2-6% tolerance band. Whether household expectations have become similarly anchored remains an open question. Using 65 waves of the Reserve Bank of India's Inflation Expectations Survey of Households (IESH)-approximately 335,000 responses collected between 2014 and 2025-we examine how broadly anchoring extends across Indian households. Around three salient shocks, one-year-ahead expectations moved in the direction the shock would predict: they fell by about 2 percentage points after demonetisation, rose by 1.8 points after the COVID-19 lockdown, and moved little on impact after the Russia-Ukraine war. Beneath the aggregate, expectations differ systematically across demographic groups. Older respondents, retired individuals, and residents of Tier 1 cities report higher expected inflation, as do women at the one-year horizon. The gap between the demographic profile with the highest predicted expectations and the profile with the lowest is on the order of two percentage points, half the width of the tolerance band. The pandemic reshuffled this heterogeneity, narrowing age differences and widening occupational ones. These patterns suggest that anchoring is better read as a distribution than as a single measure, and they help identify the household groups for whom targeted central-bank communication may yield the greatest gains.

The persistent gap between household inflation expectations and measured inflation complicates the interpretation of surveys for monetary policy. Using the Reserve Bank of India’s Inflation Expectations Survey of Households microdata for 2014–25, this article examines forecast bias, reporting patterns and the consequences of filtering responses. Households overpredict urban consumer price inflation by about five percentage points on average, with greater overprediction in the post-pandemic sample, particularly at the one-year horizon. Calendar patterns explain little individual variation, while rounding and the open-ended upper response category shape the reported distribution. Filters based on proximity between current inflation perceptions and official CPI bring aggregate expectations closer to realised inflation, but exclude substantial shares of responses. The three- and four-percentage-point bands retain 47.3% and 56.5% of observations, respectively. For monetary policy, closer agreement with CPI must be weighed against whose expectations remain in the measure.

India spent nearly $72 billion on gold imports in 2025–26, despite its substantial domestic holdings. This article argues that reducing import dependence requires distinguishing gold held for consumption from gold held for investment. Jewellery’s cultural and emotional significance limits households’ willingness to release it, while investment holdings offer greater scope for substitution towards financial assets. Reviewing India’s existing gold schemes and drawing lessons from Indonesia, the article proposes a regulatory framework that makes digital gold accessible and trustworthy through clear customer rights, verified backing, secure custody and reliable withdrawals. A separate, voluntary connection to bank deposits and lending could make existing domestic gold available to jewellers. Digitalisation alone, however, need not reduce imports if providers purchase equivalent quantities of new gold to support household balances. The policy priority is therefore to build credible financial alternatives to physical gold while enabling domestic holdings to replace imported supply.

We document the structure of interest rates faced by Indian households across formal and informal credit sources using the All India Debt and Investment Survey (AIDIS) 2019. Despite decades of financial sector reform, a substantial share of household borrowing in India occurs outside the formal system, and the rates borrowers pay vary widely both across and within sectors. We find that informal rates exceed formal rates by roughly 15–20 percentage points on average, with professional moneylenders charging the highest rates. Observable borrower and loan characteristics account for less than half of the formal-informal gap, and residual gaps are largest for SC/ST households and those in the bottom asset quintiles, suggesting that unequal access, rather than unequal risk, drives much of the dispersion.